Arbor is a multi-agent framework that introduces structured tree search as a cognition layer for autonomous agents operating in large, stateful action spaces. Prior autonomous optimization systems operate on isolated targets with stateless evaluation. Arbor instead maintains an explicit search tree of scored hypotheses that serves as the shared working memory across agents, evolving with every measurement, treating failures as diagnostic signal that reshapes subsequent exploration, and expanding as prior successes shift the bottleneck distribution. We validate Arbor on full-stack LLM inference optimization, a domain where achieving peak performance has historically required coordinated effort from engineering teams across the application, framework, compiler, kernel, and hardware stack. Arbor pairs an Orchestrator agent, which drives optimization by delegating to Domain Specialists across the inference stack, with a Critic agent that safeguards stability through root-cause analysis, introspection, and measurement validation -- a checks-and-balances architecture where neither agent can unilaterally drive the system. Agent capabilities are decomposed into hard skills (domain expertise) and soft skills (coordination protocols that determine how contributions compose), enabling fully autonomous multi-day campaigns. Arbor achieves up to 193% inference throughput-latency Pareto improvement over vendor-optimized baselines, while a single agent without the harness plateaus at +33% throughput improvement and crashes irrecoverably within hours. Arbor generalizes to multiple generations of hardware platform, and run-to-run variance is within 2 percentage points demonstrating that the method is hardware-agnostic and reproducible.

Traditionally, decision support studies how humans use machine learning models to make better decisions. In modern agentic systems, this division of roles is increasingly reversed: AI agents act on behalf of users, while humans and tools becomes support mechanisms around them. This role reversal brings reliability concerns to the forefront, since agentic errors can be consequential and agent behavior must remain aligned with human goals and constraints. Departing from the classical view of decision support, we revisit its two basic principles, the cost--value tradeoff of seeking support and the role of uncertainty quantification, in a setting where AI agents are the central actors. We propose a framework for strategic decision support for AI agents through an optimization problem that minimizes support usage subject to controlling a counterfactual missed-support error: the probability that the agent acts alone on instances where support would have materially improved its output. At the population level, we show that the optimal policy is a threshold rule on the value of support. Building on this structure, we develop an online algorithm that adaptively thresholds such a score and uses randomized exploration to control missed-support error without distributional assumptions. We further introduce a calibration-on-the-fly method that reduces unnecessary support calls online. We instantiate this framework across diverse scenarios, including information gathering, human--AI collaboration, and tool use, showing how each can be modeled through the same strategic decision-support lens. Experiments across these settings show that our method reliably controls the target error while substantially reducing support usage in practice.

Human mobility data is important for transportation, urban planning, and epidemic control, but large-scale trajectory collection is often costly and privacy-constrained, motivating realistic synthetic trajectory generation. Existing LLM-based generators typically rely on either prompt engineering, which preserves zero-shot reasoning but lacks fine-grained spatiotemporal grounding, or trajectory-level fine-tuning, which improves statistical precision but incurs substantial computational cost and may weaken general reasoning. We propose TrajGenAgent, a semantic-aware hierarchical LLM-agent framework for human mobility trajectory generation without model fine-tuning. TrajGenAgent uses a two-stage orchestrator-worker design: an LLM first synthesizes an individual- and weekday-conditioned activity chain from historical evidence via in-context learning, and a deterministic workflow then grounds each activity into a complete visit using personalized POI retrieval, distance-aware location selection, kinematics-aware travel-time propagation, and LLM-based duration estimation. To evaluate realism beyond aggregate spatiotemporal statistics, we introduce an anomaly-detection-based evaluation framework using two complementary detectors to assess behavioral and semantic plausibility. Experiments on benchmark and large-scale simulation datasets show that TrajGenAgent improves spatiotemporal fidelity, semantic coherence, and individual-specific behavioral realism over representative neural and LLM-based baselines, while avoiding parameter updates.

Compact language models (LMs) reduce cost, latency, and deployment risk for tool agents. Yet MCP-style tool use requires more than isolated function calling: an agent must discover tools from live catalogs, satisfy schemas, preserve dependencies across intermediate outputs, and ground final responses in executed evidence. Small planners often generate plausible workflow graphs that fail under tool resolution, parameter validation, dependency tracking, or execution. We argue that this failure mode is poorly handled by small-corpus distillation. A few hundred teacher traces can teach workflow format, but rarely cover the recovery behavior needed to repair failed plans over changing tool catalogs. We introduce Evoflux, an inference-time evolutionary search method that treats compact tool use as the repair of executable tool workflows. It evolves typed workflow graphs through structured edits, execution feedback, adaptive intensity, meta-guided redesign, and diversity pruning. On held-out MCP-Bench tasks spanning live MCP servers and 250 tools, Evoflux raises execution feasibility from roughly 3% to 17-24% across small planners. In contrast, SFT and SFT+DPO on the same search-mined data match, underperform, or collapse below zero-shot performance; ReAct reaches higher peaks, but with higher variance and token cost. These results show that execution-grounded search is more reliable under scarce teacher-trace budgets.

Rapid advances have been made in developing general-purpose embodied agent in environments like Minecraft through the adoption of LLM-augmented hierarchical approaches. Despite their promise, low-level controllers often become performance bottlenecks due to repeated execution failures. We argue that a key limitation is not only the lack of episodic memory, but also the decoupling of \textit{what-where-when} memory from \textit{which-why} reasoning. To address this, we propose \textbf{WISE} (Which-Why Informed Semantic Explorer), a long-horizon agent framework with an enhanced low-level controller equipped with a Causal Event Graph that augments episodic memory with explicit causal structure linking observations to task relevance. Unlike prior work such as MrSteve, which relies on feature similarity for retrieval, WISE enables robust recall under viewpoint changes and supports opportunistic task reordering through causal reasoning. Building on this memory, we propose an Opportunistic Task Scheduler that dynamically re-prioritizes subtasks when causally relevant opportunities are detected. We further equip WISE with a multi-scale progressive exploration strategy to provide spatially comprehensive observations for downstream reasoning. Experiments show that WISE largely improves task success and efficiency on long-horizon sparse tasks, particularly in settings requiring adaptive decision-making.

Large language models are increasingly deployed as agents for long-horizon tasks, yet their performance is shaped not only by model capability and environment design, but also by the harness that mediates agent--environment interaction. Existing harnesses are largely manually engineered, making them difficult to scale as trajectories grow longer and interactions become more complex. In this work, we ask whether harness can be generated by a learnable plug-in module that can be trained in an end-to-end fashion. We introduce HarnessBridge, a lightweight learnable harness controller that parameterizes the agent--environment interface as a bidirectional projection. HarnessBridge learns two bidirectional projections: observation projection, which distills raw trajectories into compact, decision-relevant states, and action projection, which converts proposed actions into executable transitions or trajectory-grounded rejections. We train HarnessBridge on a harness supervision dataset via unified instruction tuning. On Terminal-Bench~2.0 and SWE-bench Verified, HarnessBridge matches or surpasses strong specialized harnesses while substantially reducing token usage and trajectory length, and generalizes from smaller generators to larger commercial models.

We present a modular two-agent simulation framework for evaluating conversational shopping assistant architectures. An independent buyer agent, configured with personas, missions, and patience levels, is paired with an interchangeable responder that integrates with a real e-commerce search API. Holding the buyer constant across experiments enables controlled comparison of responder designs on identical scenarios. Using 2011 conversations across 14 persona buckets, we establish four empirical findings. First, rolling-window memory outperforms intent-extraction memory on all quality metrics while being 35% faster per query. Second, illustrating rapid evidence-driven iteration, a systematic failure analysis of a responder version enables targeted fixes that reduce failure and near-failure rates by 62% across the full dataset. Third, swapping the responder LLM backbone from Gemini~2.5 to Llama~3.3~70B costs 0.16--0.45 points despite identical architecture. Finally, we document systematic philosophical disagreement between frontier LLM judges: Gemini rewards process correctness while Claude demands concrete outcomes, despite using the same evaluation prompt.

Long-running LLM agents accumulate interaction histories far larger than any context window, forcing a standing decision: what to encode deeply, what to forget, and what to retrieve under a fixed memory budget. Production systems answer with semantic similarity or recency -- both mis-specified for the forgetting decision, which is made at consolidation time before the future query is known. We propose a multi-factor memory value function V(m)=\sum_i w_i f_i(m) over seven interpretable factors (emotional intensity, goal relevance, value alignment, self/user relevance, task utility, reliability, and usage history) drawn from cognitive psychology, whose weights are learned from a downstream objective by a gradient-free optimiser, and whose single scalar uniformly controls encoding depth, forget risk, and retrieval rank. We make a methodological point: on LongMemEval, scoring goal relevance against the held-out evaluation question saturates gold-evidence retention at \approx 0.98 -- this measures retrieval, not forgetting. In the realistic blind regime, a learned multi-factor value retains 0.770 \pm 0.011 of gold evidence across 479 usable cases, versus 0.657 for uniform weights, 0.518 for the best single factor, and 0.368 for recency; every paired gap's 95% bootstrap CI is above zero, and a neural network over the same factors ties the linear model. The learned weights are interpretable -- reliability, emotional intensity, and self/user relevance dominate, while query-time goal similarity is correctly down-weighted for the forgetting decision. A controlled synthetic task with planted confounds confirms the learner recovers a separating weighting (1.00 retention) where uniform weighting fails (0.62). The substrate is open-source; all experiments run on a single CPU with no API calls.

As LLM agents proliferate in prediction markets and collective decision-making, they risk a cognitive monoculture: agents built on shared foundation models produce correlated forecasts, and recent measurement finds frontier-model errors correlated at r ~ 0.77. We ask whether human cognitive diversity can be recovered from behavior and transferred to LLM agents. Nous extracts a structured eight-dimension behavioral profile from real Polymarket trading activity and injects it into agents through prompts. Our central finding is a dissociation between the two halves of that pipeline. Extraction works, partially: across 100 wallets, 8 of 14 parameters are temporally stable (split-half ICC >= 0.5, bootstrap CI lower bound > 0.3; contrarian score reaches ICC ~ 0.9); wallets are identifiable from their profiles well above chance (top-1 retrieval 17-22% vs. 1% chance); and two of four pre-specified dimensions rank-correlate with future realized profit out-of-sample, though the correlations do not survive behavioral-confound controls. Prompt-level injection does not measurably transmit it: on a semantic embedding metric, structured injection shows no significant advantage over a length-matched control on any model, and the diversity it induces neither reduces ensemble error correlation nor improves Brier score -- a null that persists across exploratory checks on sampling temperature, profile diversity, and question difficulty. Measuring the prompts themselves locates the compression before the model: the structure-to-narrative translator emits near-uniform prompts whose spread does not track profile spread. We position Nous as measuring the cognitive-monoculture problem and the limits of a prompt-level remedy, motivating deeper, below-the-prompt injection (fine-tuning, activation steering). Code, frozen profiles, prompts, and model outputs: https://github.com/WillChienT/nous-paper

Large language models (LLMs) are increasingly used as interactive assistants for technical problem solving. However, when users provide incomplete descriptions or plausible but unverified explanations, LLMs may prematurely align with these assumptions and propose solutions before collecting sufficient evidence. We refer to this behavior as user-driven sycophancy: the tendency of an LLM to reinforce a user-provided hypothesis instead of testing alternative explanations. This paper introduces LLM-as-an-Investigator, an evidence-first agentic AI methodology for robust problem diagnosis. The approach is implemented through a Solution Investigator Agent, which estimates the ambiguity of an initial problem description, generates candidate hypotheses, asks targeted clarification questions, and updates hypothesis probabilities after each answer. Rather than producing an immediate response, the agent continues the investigation until the evidence makes one candidate explanation stronger than the alternatives. To evaluate the approach, we build a benchmark from solved technical forum threads in mechanical, electrical, and hydraulic domains. We use a three-agent evaluation pipeline in which a Problem-Solution Extractor Agent converts solved threads into structured cases, a Ground-Truth Evaluator Agent simulates the user while hiding the known solution, and the tested assistant attempts to recover the solution through dialogue. The experiments compare standard assistants, reasoning-oriented LLMs, and the proposed investigator-based model across LLM backbones. In addition to diagnostic accuracy, we analyze how standard assistants follow misleading user hypotheses in diagnostic cases. The results show that the proposed approach identifies the problem more accurately than direct prompting and reasoning-only baselines, while its evidence-first protocol helps reduce user-induced conversational bias.

Recent approaches combining Large Language Models (LLMs) with retrieval-augmented reasoning have shown promise for automated fact verification. To process complex claims, these verification pipelines typically execute multi-stage workflows that coordinate tightly coupled modules, including claim decomposition, evidence gathering, and verdict prediction. However, existing methods optimize individual stages in isolation or rely on fixed heuristics, which limits adaptive coordination among stages and can lead to suboptimal outcomes. In this work, we propose ProFact, an agentic reinforcement learning framework for end-to-end optimization of multi-stage fact verification trajectories. ProFact trains a unified policy to coordinate claim decomposition, evidence seeking, answer generation, and verdict prediction. To address the sparse and delayed supervision provided by final veracity labels, ProFact introduces process-aware rewards that provide stage-level learning signals throughout the verification process. Empirical evaluation shows that ProFact consistently outperforms strong baselines in both verification performance and inference efficiency. These results highlight the effectiveness of process-aware trajectory optimization for multi-stage fact verification.

Right now, across the world, AI agents are repeating the same absurd act: to read one document, they each recompute it from scratch. Every agent re-runs prefill, the most compute-intensive step a large model takes, over identical text, only to rebuild a key-value (KV) cache identical to the one the agent before it just built. The same answer, computed a million times. We make a proposal that is almost offensively simple: compute it once. Let a publisher precompute a document's KV cache, and let every other agent buy the right to load it and skip prefill. It works, and it is token-exact: loading a precomputed KV and continuing matches prefilling from scratch (24/24 greedy tokens, and at the logits level), with no accuracy cost. On Qwen3-4B, reuse is 9-50x cheaper in compute than prefill, and the gap widens with length (prefill's attention scales with L^2), so a single reuse already pays it back. Then the part that matters: where the KV lives. Shipping it fails, because KV is nearly incompressible, so per-load egress costs more than the prefill it saves. Hosting it provider-side, exactly as production prompt-caching works, removes egress entirely. The size of the prize is set by our measured compute saving: serving one hot 3774-token document to 80M agents costs ~$1.5M to re-prefill but only ~$0.03M of reuse compute (49.7x less). The 0.1x cache-read tariff APIs charge passes a 10x discount to users while sitting inside this measured envelope, so the 10x is a floor that the measured ~50x compute saving clears, and the gap to the physical ~50x is provider margin: millions of dollars per popular document. We frame the resulting agent-native prefill CDN and leave lossless KV compression and a cross-party payment layer as the open problems.

Computer-Aided Design is pivotal in modern manufacturing, yet existing automated methods predominantly rely on open-loop, one-shot generation, creating a mismatch with iterative real-world practices. In this paper, we present IterCAD, a unified multimodal agent framework for closed-loop, interactive CAD generation and editing. We formulate the task as a multi-turn interaction between a multimodal agent and an executable CAD sandbox, covering three tasks: Drawing-to-Code, Text-to-Code, and Interactive Editing. To support this, we develop a data synthesis pipeline incorporating advanced industrial manufacturing features to generate standard-compliant multi-view engineering drawings, complex code-editing tasks, and high-fidelity interaction trajectories. We optimize the agent via progressive SFT followed by geometry-aware reinforcement learning with viable-prefix masking to enhance code executability and geometric fidelity. Finally, we introduce the IterCAD-Bench evaluation suite and propose the Chamfer Distance Tolerance-Recall (CD-TR) curve alongside its AUC-TR metric, establishing a survivor-bias-free standard that unifies code validity and geometric precision. Extensive experiments demonstrate that IterCAD achieves highly competitive performance across multiple benchmarks, significantly outperforming existing approaches in both code executability and geometric precision, while exhibiting superior capabilities in closed-loop iterative refinement.

Training interactive web agents through imitation learning from expert trajectories has emerged as a highly effective approach. However, determining the optimal timing for expert intervention presents a critical challenge in this context. Delayed intervention often leads to the accumulation of early-stage errors, pushing the page state into an irrecoverable regime. Conversely, premature or excessive intervention causes the agent to become overly reliant on expert policies, trapping the model in local optima characterized by a single, rigid trajectory. We propose Speculative Rollback Correction (SRC), a branch-level imitation framework for resettable agent environments. Instead of requesting teacher labels at every visited state or correcting only after a completed trajectory, SRC uses fixed-horizon branch review: the student executes a short speculative segment before teacher review, and the teacher localizes the first harmful deviation only when local progress breaks. Rollback preserves useful prefixes, while successful rollouts are filtered by a hard verifier and retained in a lightweight quality-diversity archive. The resulting data supports next-action supervised fine-tuning on both localized corrections and verifier-passing trajectories. On WebArena-Infinity, SRC collects 977 verifier-passing trajectories and 9,183 next-action examples; fixed-horizon review improves the recovery-versus-query tradeoff over step-level review while retaining verifier-passing solution variants. Code is available at https://github.com/LongkunHao/SRC_gui_agent.

Long-horizon tool-use reinforcement learning can learn from outcome verification, but its trajectory-level advantage is broadcast across many reasoning, API, and answer tokens. Self-distillation promises a denser signal by reusing a policy's own rollouts or a privileged teacher. We show, however, that direct token-level self-distillation can silently destroy tool use: it rehearses teacher behavior without knowing which actions the verifier rewards, so useful skills and harmful shortcuts are amplified together. We introduce Sibling-Guided Credit Distillation (SGCD), which uses distillation for credit assignment rather than as a competing actor loss. Dynamic sampling produces mixed successful and failed sibling rollouts; an external LLM summarizes their contrast into a training-only stepwise credit reference; dense teacher/student divergence drives credit reassignment; and bounded detached credit weights reshape GRPO token advantages. The deployed student sees no external LLM, sibling evidence, or oracle. Across AppWorld and $τ^3$-airline, SGCD improves over matched GRPO comparators: AppWorld TGC $42.9 \to 45.6$ on test_normal and $24.7 \to 27.0$ on test_challenge, and $τ^3$-airline pass@1 $0.583 \to 0.602$.

While MPC effectively handles structured, diverse, and low-level specifications, it lacks the capability to dynamically incorporate high-level contextual information such as social norms, user intent, or natural language instructions. To address this limitation, this manuscript introduces an agentic MPC framework that enables context-aware, semantically adaptive control synthesis by integrating with large language model-based agents. The agent interprets heterogeneous inputs, including natural language messages, environmental observations, and external knowledge, to resynthesize the control specifications. The effectiveness of the framework is demonstrated in an autonomous driving scenario, where the system aligns with personal preferences or responds to social situations such as emergency vehicle yielding.

While recent vision-language models (VLMs) demonstrate strong multimodal understanding, they remain limited in spatial reasoning tasks that require active evidence acquisition and multi-step visual interaction. This limitation suggests that relying solely on implicit visual representations from vision encoders is insufficient for recovering fine-grained spatial evidence. We introduce PERception-Interaction-reason Agent (PERIA), a tool-augmented visual agent for spatial reasoning tasks across map reasoning, visual probing, and vision reconstruction. PERIA uses two lightweight tool families: vision perception tools for exposing textual, symbolic, and spatial evidence, and vision interaction tools for manipulating visual context, tracing paths, and verifying spatial relations. To train PERIA, we develop a unified recipe that combines supervised tool-use trajectory synthesis, composite rewards, and Observation-Relaxed Group-in-Group Policy Optimization (OR-GIGPO) for effective multi-tool behavior. Experiments on 13 benchmarks from 8 datasets show that PERIA-8B improves over the Qwen3-8B backbone by 10.0% on in-distribution benchmarks and 4.4% on out-of-distribution benchmarks, while outperforming previous state-of-the-art baselines of similar size by 7.0%-14.8%. It also achieves performance comparable to much larger models such as Qwen3-VL-235B-A22B-Thinking and GPT-5, demonstrating the effectiveness of PERIA in enhancing spatial reasoning capabilities.

Existing computer-use agents remain fundamentally limited in professional software manipulation: GUI-based agents suffer from fragile visual grounding and long-horizon error accumulation, while API-basedapproaches struggle with heterogeneous protocols and inaccessible commercial interfaces. In this work,we identify the Component Object Model (COM) as a unified executable abstraction, proposing COM-as-Action: a new paradigm that reframes professional software interaction as deterministic program synthesisrather than sequential visual control. To validate this paradigm in the most demanding environments, weintroduce ComCADBench, the first benchmark for agents operating real industrial CAD software. Ourexperiments reveal a substantial paradigm gap: frontier proprietary models achieve near-zero successunder GUI-based interaction, whereas COM-based execution yields substantial immediate gains. Tobridge the remaining gap between syntactic correctness and geometric accuracy, we develop ComActor, aself-correcting agent trained through a progressive three-stage framework, alongside ComForge, a scalableplatform for large-scale training in Windows containers. Extensive experiments show that ComActorachieves state-of-the-art performance on ComCADBench, with strong resilience in long-horizon taskswhere baselines collapse, and generalizes to external CAD benchmark.

Spatial reasoning, the ability to determine where objects are, how they relate, and how they move in 3D, remains a fundamental challenge for vision-language models (VLMs). Tool-augmented agents attempt to address this by augmenting VLMs with specialist perception modules, yet their effectiveness is bounded by the action interface through which those tools are invoked. In this work, we study how the design of this interface shapes the agent's capacity for open-ended spatial reasoning. Existing spatial agents either employ single-pass code execution, which commits to a full analysis strategy before any intermediate result is observed, or rely on a structured tool-call interface that often offers less flexibility for freely composing operations or tailoring the analysis to each task. Both designs offer limited flexibility for open-ended, complex 3D/4D spatial reasoning. We therefore propose SpatialClaw, a training-free framework for spatial reasoning that adopts code as the action interface. SpatialClaw maintains a stateful Python kernel pre-loaded with input frames and a suite of perception and geometry primitives, letting a VLM-backed agent write one executable cell per step conditioned on all prior outputs, enabling the agent to flexibly compose and manipulate perception results and adapt its analysis to both intermediate text and visual observations and the demands of each problem. Evaluated across 20 spatial reasoning benchmarks spanning a broad range of static and dynamic 3D/4D spatial reasoning tasks, SpatialClaw achieves 59.9% average accuracy, outperforming the recent spatial agent by +11.2 points, with consistent gains across six VLM backbones from two model families without any benchmark- or model-specific adaptation.

As agents grow more capable, legal-domain LLM agents promise to turn document-heavy matters into reviewable work products -- yet reliable deployment faces three obstacles: no large-scale evidence on how today's strongest model-and-harness combinations behave on end-to-end legal matters; no agent architecture adapted to the legal vertical, only general-purpose harnesses; and, in a setting that keeps shifting with new facts, authorities, and deadlines, no mechanism for systems to learn from their own outcomes. We address each. A large-scale empirical study on Harvey LAB -- $12{,}510$ agent trajectories -- shows that even frontier agents remain far from completing matters in a single pass: per-criterion accuracy climbs with stronger models while strict matter completion stalls. We then introduce \textsc{Parthenon}, a self-evolving legal-agent framework that factors Model, Harness, Agent roles, legal Knowledge, deterministic Tools, and procedural Skills into auditable surfaces for source traceability, date and number grounding, deliverable compliance, and issue closure. Finally, an anti-leakage learning loop converts scored failures into task-agnostic edits to skills, tools, and knowledge, letting the system improve with experience -- as a firm refines its checklists and playbooks after each matter -- without touching model weights. Across our large-scale empirical analysis, \textsc{Parthenon} substantially improves the performance of state-of-the-art models and harnesses on legal-matter tasks.

Continual reinforcement learning challenges agents to acquire new skills while retaining previously learned ones with the goal of improving performance in both past and future tasks. Most existing approaches rely on model-free methods with replay buffers to mitigate catastrophic forgetting; however, these solutions often face significant scalability challenges due to large memory demands. Drawing inspiration from neuroscience, where the brain replays experiences to a predictive World Model rather than directly to the policy, we present ARROW (Augmented Replay for RObust World models), a model-based continual RL algorithm that extends DreamerV3 with a memory-efficient, distribution-matching replay buffer. Unlike standard fixed-size FIFO buffers, ARROW maintains two complementary buffers: a short-term buffer for recent experiences and a long-term buffer that preserves task diversity through intelligent sampling. We evaluate ARROW on two challenging continual RL settings: Tasks without shared structure (Atari), and tasks with shared structure, where knowledge transfer is possible (Procgen CoinRun variants). Compared to model-free and model-based baselines with replay buffers of the same-size, ARROW demonstrates substantially less forgetting on tasks without shared structure, while maintaining comparable forward transfer. Our findings highlight the potential of model-based RL and bio-inspired approaches for continual reinforcement learning, warranting further research.

Reinforcement learning (RL) in robotics is often limited by the cost and risk of data collection, motivating experience transfer from a source task to a target task. Offline-to-online RL leverages prior data but typically assumes a given fixed dataset and does not address how to generate reliable data for transfer. We propose World Model-Based Experience Transfer (WOMBET), a framework that jointly generates and utilizes prior data. WOMBET learns a world model in the source task and generates offline data via uncertainty-penalized planning, followed by filtering trajectories with high return and low epistemic uncertainty. It then performs online fine-tuning in the target task using adaptive sampling between offline and online data, enabling a stable transition from prior-driven initialization to task-specific adaptation. We show that the uncertainty-penalized objective provides a lower bound on the true return and derive a finite-sample error decomposition capturing distribution mismatch and approximation error. Empirically, WOMBET improves sample efficiency and final performance over strong baselines on continuous control benchmarks, demonstrating the benefit of jointly optimizing data generation and transfer.

Inferring others' beliefs requires more than reading surface signals; it requires tracking who told them what, in what order, and how credibly. The Theory of Mind Utility (ToM-U) formalizes this epistemic state inference problem at the computational level of analysis, specifying what mentalizing computes and why without commitment to algorithmic or neural implementation. ToM-U achieves this by constructing Local Epistemic World Models (LEWMs) -- directed typed graphs that represent agents, state nodes, and the epistemic relationships among them -- and evaluating discrete candidate LEWMs against observed behavior until one achieves sufficient confidence. Five formal definitions specify the LEWM structure, agent node properties including ordered information access history, a bounded proliferation mechanism for recursive mentalizing, three inference procedures, and a residue function that captures the structured trace left by failed mentalizing attempts. ToM-U differs from Bayesian Theory of Mind and adjacent formal accounts, which presuppose rather than derive belief states, and from simulation theory and theory-theory, which lack a formal apparatus for epistemic state inference. The architecture generates directional, falsifiable predictions about mentalizing failure that follow from structural properties of the model rather than auxiliary assumptions, and positions ToM-U as a domain-agnostic mechanism upstream of goal inference and other downstream social cognitive processes.

LLM-based agents are entering regulated industries where they automate judgment intensive quality management processes. We argue that symbolic structures already embedded in these domains, including regulations, typed process models, and compliance constraints, should be treated not merely as external monitoring mechanisms but as core architectural components that shape the agent's decision-making and behavior. We propose compliance-by-construction as a complementary paradigm to guardrail-based monitoring: a structural foundation that prevents control-flow violations, while guardrails remain essential for catching semantic errors. We identify a structured set of neuro-symbolic research challenges on foundational and capability level and show that addressing them jointly enables compliance-by-construction. We call on the neuro-symbolic community to engage with regulated process automation as a high impact research domain.

Current LLM-based research agents have advanced through agent orchestration, yet largely overlook scientific knowledge orchestration. Existing works often reduce papers to abstracts, surface mentions, and flat \texttt{cites} edges, omitting key entities, claims, evidence, mechanisms, and method lineages essential for scientific reasoning. To this end, we introduce \textbf{Agents-K1}, an end-to-end knowledge orchestration pipeline that converts raw documents into agent-native scientific knowledge graphs. Agents-K1 integrates three components under a unifying theoretical foundation: a multimodal parser whose five-module schema captures entities, multimodal evidence, citations, and typed inter-entity relations across the full paper rather than abstracts alone; a 4B information-extraction backbone trained with GRPO under a rule-based reward; and a graphanything CLI, a tri-source agent interface that unifies web search, multimodal graph retrieval, and cross-document traversal. On top of this, we process 2.46 million scientific papers across six subjects to produce \textbf{Scholar-KG}, of which we release a one-million-paper subset, and the full Scholar-KG is accessible via the SCP link below. The same pipeline can be extended to general-domain corpora and to schema-conformant data synthesis. Extensive experiments demonstrate that Agents-K1 achieves superior performance in scientific information extraction, knowledge graph construction, and multi-hop scientific reasoning.

Recent large language models (LLMs) have achieved impressive reasoning milestones but continue to struggle with high computational costs, logical inconsistencies, and sharp performance degradation on high-complexity problems. While neuro-symbolic methods attempt to mitigate these issues by coupling LLMs with symbolic reasoners, existing approaches typically rely on monotonic logics (e.g., SMT) that cannot represent defeasible reasoning -- essential components of human cognition. We present "LLM+ASP," a framework that translates natural language into Answer Set Programming (ASP), a nonmonotonic formalism based on stable model semantics. Unlike prior "LLM+ASP" approaches that require manually authored knowledge modules, domain-specific prompts, or evaluation restricted to single problem classes, our framework operates without any per-task engineering and applies uniformly across diverse reasoning tasks. Our system utilizes an automated self-correction loop where structured feedback from the ASP solver enables iterative refinement. Evaluating across six diverse benchmarks, we demonstrate that: (1) stable model semantics allow LLMs to naturally express default rules and exceptions, outperforming SMT-based alternatives by significant margins on nonmonotonic tasks; (2) iterative self-correction is the primary driver of performance, effectively replacing the need for handcrafted domain knowledge; (3) compact in-context reference guides substantially outperform verbose documentation, revealing a "context rot" phenomenon where excessive context hinders constraint adherence.

Active inference casts decision-making as inference, with the Expected Free Energy (EFE) unifying goal-directed and information-seeking behavior. Recent work showed that EFE minimization can be written as Variational Free Energy (VFE) minimization on a generative model augmented with epistemic priors. We prove that the VFE of the augmented model can be rewritten as the VFE of the predictive model plus explicit entropy-correction terms, making the EFE contribution transparent. We then show that proper EFE-based planning requires combining these epistemic corrections with a planning correction that turns marginal inference into policy optimization, yielding a full variational characterization of EFE-based planning. This clarifies which corrections are needed for cross-entropy planning and for full EFE-based planning. The same entropy-corrected formulation leads to a detailed message-passing scheme for EFE-based planning together with simpler ablations. Experiments on three grid-world environments show that full EFE-based planning outperforms ablations that omit either the planning correction or the epistemic corrections.

We propose KG-ER, a conceptual schema language for knowledge graphs that describes the structure of knowledge graphs independently of their representation (relational databases, property graphs, RDF) while helping to capture the semantics of the information stored in a knowledge graph.

Time Series Event Detection (TSED) aims to localize semantically meaningful events in time series data, with critical applications in high-stakes domains. Unlike statistical anomalies, events are often defined by natural-language descriptions with internal temporal-logic structures across multiple physical channels. However, in real-world settings, dense event annotations are expensive to obtain, making purely supervised learning difficult. We introduce Language-guided TSED, a setting where a model is given textual event descriptions and must ground them to intervals in multivariate signals with little or no labeled data. To address this problem, we propose Event Logic Tree (ELT), a knowledge representation framework that converts linguistic descriptions into structured temporal logic over signal primitives. Building on ELT, we present SELA, a neuro-symbolic VLM agent framework that iteratively grounds primitives from signal visualizations and composes them under ELT constraints, producing both event intervals and faithful tree-structured explanations. We further release a real-world benchmark across energy and climate domains with expert knowledge and annotations. Experiments show that SELA improves over supervised fine-tuning and existing zero/few-shot time series reasoning baselines.

Multi-agent debate (MAD) can improve large language model reasoning, but fixed debate pipelines often waste computation and can amplify correlated errors among similar agents. We propose ARMOR-MAD, a training-free heterogeneous MAD framework that treats debate as conditional computation. ARMOR-MAD combines three components: Pre-debate Agreement Routing (PAR) decides whether independently generated Round-0 answers require debate; Early Agreement Stopping Evaluator (EASE) stops debate after convergence; and Semantic Outlier Detection (SOD) down-weights abnormal final answers during aggregation. Across MATH Level 5, GSM8K, MMLU, and MMLU-Pro, ARMOR-MAD consistently improves over fixed-round heterogeneous debate with the same model pool, reaching 65.5\%, 96.5\%, 90.0\%, and 81.5\% accuracy, respectively. The results suggest that genuine model heterogeneity and agreement-based control are both important for making MAD more accurate and efficient.

Confidence is used for reliability, oversight, and a range of downstream decision tasks in Natural Language Processing (NLP), yet no existing method produces or evaluates a confidence for the output of a multiagent system. Prior work uses confidence within multiagent debate (MAD) to weight messages, trigger debate, or calibrate individual agents, but it never aggregates these into a single confidence for the system itself. We introduce three protocols that produce a final answer along with a single aggregated confidence by first transforming raw confidence signals to make them comparable across models, then combining them via soft voting or a probability fusion we call Bayesian fusion. This aggregated confidence is substantially more discriminative (AUARC) than that of the best single agent or the standard debate baselines, while correctness (F1-score) stays stable and recovers the losses MAD incurs on more ambiguous tasks. Analyzing two estimators, sequence probability and self-report, alongside parametric and non-parametric calibrators, we find that calibration improves F1 for both estimators while AUARC is less reliant on it. We evaluate six homogeneous and heterogeneous debating pairs per benchmark, across five benchmarks and four task types, spanning a range of model capabilities and sizes.

Multi-Agent Systems (MAS) built on Large Language Models (LLMs) require effective orchestration to coordinate specialized agents, yet training such orchestrators is hindered by limited supervision and high computational cost. We propose Orchestration Reward Modeling (OrchRM), a self-supervised framework for evaluating orchestration quality without human annotations. OrchRM leverages intermediate artifacts from multi-agent executions to construct win-lose pairs for Bradley-Terry reward model training. Unlike existing MAS test-time scaling and orchestrator training frameworks that rely on costly sub-agent rollouts, OrchRM operates directly at the orchestration level, enabling efficient and high-performing reward-guided orchestrator training and MAS test-time scaling. OrchRM improves training efficiency by up to 10x in token usage while improving MAS test-time scaling performance by up to 8% in accuracy. These gains consistently transfer across multiple domains, including mathematical reasoning, web-based question answering, and multi-hop reasoning, demonstrating orchestration-level reward modeling as a scalable direction for robust multi-agent orchestration. Code will be available at https://github.com/Wang-ML-Lab/OrchRM.

Dispatch in three-sided marketplaces provides a natural setting for reinforcement learning from world feedback: decisions are evaluated by delayed operational outcomes such as delivery speed, courier utilization, and merchant congestion. We present a deployed reinforcement learning system at DoorDash that adapts dispatch objective weights in a large-scale food-delivery marketplace using delayed signals. Rather than replacing the combinatorial assignment optimizer, a store-level policy learned from logged marketplace data selects a discrete multiplier that shifts the dispatch optimizer's tradeoff between delivery quality and batching efficiency. This interface enables offline policy learning under noisy, delayed, and coupled feedback while preserving production feasibility constraints and operational safeguards. We train a shared value function using centralized offline data and decentralized store-level execution, with Double Q-learning targets and a conservative regularizer to reduce out-of-distribution value overestimation. In a production switchback experiment, the offline-trained policy increases batching and reduces courier-side time costs without degrading customer-facing delivery quality. Results illustrate how world feedback from a live economic and logistics system can be used to safely adapt decision policies online.

LLM-based multi-agent systems (MAS) solve complex tasks through inter-agent collaboration, but their communication-driven nature also allows security risks to spread across agents and trigger system-wide failures. Existing MAS defenses mainly follow a reactive paradigm after execution by detecting and isolating harmful agents, which may cause irreversible damage and degrade collaborative utility. To address this, we propose a proactive defense framework for MAS security, namely a Simulation-aware Interception Guard (SAIGuard). SAIGuard performs communication-state simulation over the MAS interaction graph, estimates the impact of incoming messages on local agent states and the global MAS state, and detects risky messages via reconstruction deviations from benign communication patterns. Instead of isolating agents, SAIGuard sanitizes or regenerates suspicious messages before it propagation into system. Experiments across diverse topologies and attack scenarios show that SAIGuard reduces attack success rates while maintaining MAS utility, outperforming reactive defenses.

The rapid emergence of autonomous AI agents is transforming artificial intelligence from isolated model inference into distributed systems of reasoning, communication, and action. This paper develops the vision of the Internet of Agentic AI (IoAI): an open ecosystem in which heterogeneous agents discover one another, negotiate responsibilities, exchange context, invoke tools, and execute workflows across cloud, edge, device, organizational, and cyber-physical environments. We synthesize foundations from single-agent agentic AI, multi-agent systems, distributed computing, communication networks, game theory, and security engineering to characterize the architectures and mechanisms required for scalable agent ecosystems. The paper examines agent deployment models, workflow lifecycles, communication protocols, interoperability layers, resource-management challenges, and trust architectures, with case studies in adaptive manufacturing and distributed operational coordination. The resulting framework highlights the central research challenges of controlled emergence, semantic interoperability, secure identity, incentive-compatible coordination, resource-aware orchestration, and governance for large-scale networks of autonomous agents.

Collaborative multi-agent large language models (LLMs) can solve complex reasoning tasks by decomposing roles, but reinforcement learning for such systems is limited by credit assignment: shared terminal rewards obscure individual contributions and can encourage free-riding. We introduce two optimizer-agnostic credit assignment methods for converting joint outcomes into agent-specific learning signals. Counterfactual Credit for Policy Optimization (CCPO) estimates an agent's marginal contribution by comparing the realized joint outcome with a counterfactual outcome where that agent is removed. Self-Evaluated Credit for Policy Optimization (SEPO) uses constrained self- and peer-evaluations as a verifier-anchored credit signal while keeping the external task outcome dominant. Both operate at the reward-construction layer rather than as policy optimizers, producing role-specific rewards or advantages for GRPO, GSPO, or REINFORCE++. We instantiate these credit signals in a sequential Think--Solve setting and evaluate them on mathematical reasoning benchmarks. Results show that explicit credit assignment often improves dual-agent reasoning, especially on MATH500 and several out-of-distribution settings, while gains vary across models and datasets. Our code is available at: https://github.com/bhai114/ccpo.

We investigate the belief revision problem in epistemic planning, i.e., what will be the beliefs of all agents in a multi-agent system after an agent gains the belief in some state property. Based on the standard representation in epistemic planning of agents' beliefs via a single multi-agent Kripke model, we generalize the classical AGM belief revision postulates to the multi-agent setting, with the aim to provide a formal framework for evaluating dynamic epistemic reasoning frameworks in which the beliefs of all agents as the result of actions are computed. As an example of a simple operator that satisfies all of the generalized AGM postulates, we present generalized full-meet multi-agent belief revision. We moreover define a generalization of the standard postulates for iterated revision, present a more sophisticated, event model based revision operator, and discuss the potential issues in defining an epistemic operator on Kripke models that can satisfy all of the generalized postulates for iterated multi-agent belief revision.

Recent evidence, both in the lab and in the wild, suggests that the use of generative artificial intelligence reduces the diversity of content produced. The use of the same or similar AI models appears to lead to more homogeneous behavior. Our work begins with the observation that there is a force pushing in the opposite direction: competition. When producers compete with one another (e.g., for customers or attention), they are incentivized to create novel or unique content. We explore the impact competition has on both content diversity and overall social welfare. Through a formal game-theoretic model, we show that competitive markets select for diverse AI models, mitigating monoculture. We further show that a generative AI model that performs well in isolation (i.e., according to a benchmark) may fail to provide value in a competitive market. Our results highlight the importance of evaluating generative AI models across the breadth of their output distributions, particularly when they will be deployed in competitive environments. We validate our results empirically by using language models to play Scattergories, a word game in which players are rewarded for answers that are both correct and unique. Overall, our results suggest that homogenization due to generative AI is unlikely to persist in competitive markets, and instead, competition in downstream markets may drive diversification in AI model development.

Renewable energy is essential for meeting future energy demands; however, solar energy generation, which occurs only during daylight hours often does not align with household consumption patterns. Appliances such as cookers, washing machines, and dryers are typically operated according to user preferred schedules rather than solar energy availability, creating a scheduling optimization problem. The objective is to determine optimal appliance start times to maximize renewable energy utilization while minimizing user inconvenience and adhering to system constraints. This paper presents a metaheuristic approach using Iterated Local Search (ILS) and Simulated Annealing (SA) to optimize appliance start times, while considering appliance operating durations, power consumption, inverter limit, battery state of charge constraints, and solar generation forecasts. Unlike most existing work, the scheduling is extended beyond a single day to accommodate unfinished tasks from previous days (spillover), ensuring operational continuity and enabling sequential operation across multiple days. Experimental results show that the sequential multi-day scheduling framework effectively manages system constraints while ensuring user convenience under exclusive solar generation. These findings also open opportunities for future research on multi-objective trade-offs between investment in equipment of various sizes, return on that investment, and user satisfaction.

Rapidly expanding low Earth orbit satellite constellations are placing increasing demands on terrestrial ground networks, motivating the development of more efficient ground station network designs. Current approaches select sites from predefined locations, limiting optimization to existing infrastructure and constraining performance. In contrast, free-placement optimization operates over a continuous spatial domain on Earth, broadening the search space and allowing higher-throughput configurations at the cost of potentially requiring new infrastructure deployment. In this work, we introduce SCORE (Sequential Cyclic Optimization via Refinement & Evaluation), a two-stage free-placement method for ground station design. SCORE combines sequential coordinate selection with cyclic refinement to manage high-dimensionality, non-convexity, and local minima that challenge global optimizers. We benchmark SCORE against one-shot methods such as differential evolution (DE) and integer programming approaches using locations from Kongsberg Satellite Services and the World Teleport Association. Tests across two commercial Earth observation constellations (Capella Space and ICEYE) and one synthetic Walker-Star constellation show that SCORE requires up to 5x fewer function evaluations to converge relative to DE while improving downlink throughput by up to 13%. Compared to fixed-site methods, unconstrained SCORE achieves up to 15% greater total downlink, establishing a strong empirical performance benchmark for flexible placement; infrastructure-constrained SCORE retains over 92% of this gain while restricting placement to within proximity of existing fiber and power infrastructure. We also explore trade-offs between expanding existing stations and deploying new sites, informing future ground network design for operational constellations.

Modern Lean theorem provers achieve strong performance only with substantial training and inference compute, driven in part by scarce verified proof data and the long reasoning traces of formal proof search, making both supervised fine-tuning (SFT) and sampling expensive. We introduce Pythagoras-Prover, a compute-efficient open-source family of Lean theorem provers built for practical compute budgets. The family spans two generation paradigms: autoregressive models at 4B and 32B parameters, and a first proof-of-concept diffusion-based prover (4B) that iteratively refines Lean proofs at inference time. For training efficiency, we build a Lean-verified corpus stratified into easy, medium, and hard problems for curriculum SFT, so models acquire proof skills progressively from shorter, simpler proofs to longer, harder ones. During SFT, a dynamic proof-reasoning filtering scheme preserves informative proof traces while keeping each instance within an 8k-token context budget. We also introduce Augmented Lean Formalisation (ALF), which expands scarce verified corpora into variants of formal statements, populated via self-distillation for extra training signal without formally verifying every mutated instance. By perturbing known problems while preserving their formal character, ALF reduces reliance on any statement's surface form. Empirically, Pythagoras-Prover-4B surpasses DeepSeek-Prover-V2-671B at pass@32 on MiniF2F-Test (86.1% vs 82.4%) with ~167x fewer parameters, while Pythagoras-Prover-32B sets the open-source state of the art at 93.0% on MiniF2F-Test and solves 93 of 672 PutnamBench problems. We release MiniF2F-ALF, an ALF-mutated contamination-sensitive benchmark on which every evaluated model loses accuracy; here our 32B remains strongest and our 4B matches the prior state of the art, Goedel-Prover-V2-32B.

In Low-Rank Adaptation (LoRA), the scaling factor $α$ is often treated as a mere complement to the learning rate, yet its role in optimization remains poorly understood. In this paper, we reveal that the scaling factor $α$ and the learning rate function differently, with $α$ emerging as the dominant driver of effective optimization, delivering gains that cannot be replicated by learning rate scaling alone. Through the synergy of extensive empirical analysis and a theoretical Signal-Drift framework, we uncover three findings into LoRA's scaling mechanism: First, LoRA's spectral suppression smooths the optimization landscape, rendering standard hyperparameters overly conservative and creating an optimization gap. Second, when leveraging this smoothness to accelerate convergence, $α$ outperforms the learning rate by amplifying the task signal without increasing the drift ratio. Third, the optimal scaling factor follows a sublinear relationship with the rank, well characterized by a square-root law with an unexpectedly large coefficient, revealing the insufficient scaling of existing rank-tied heuristics. Based on these insights, we propose LoRA-$α$, a minimalist framework that restores $α$ to its principled regime, making LoRA compatible with standard small learning rates. Extensive evaluations across diverse tasks demonstrate that LoRA-$α$ consistently improves performance while streamlining hyperparameter search, unleashing the learning potential of LoRA.

Parallel test-time scaling samples many reasoning traces and majority-votes their answers, improving LLM accuracy but requiring traces to run to completion, incurring substantial computational overhead. We observe that probing partial traces at intermediate checkpoints can extract current answers without disrupting generation, revealing an evolving aggregate vote. Based on this observation, we introduce MARS, a margin-adversarial stopping rule that estimates which active traces are likely to change their answers and stops once the leader remains safe under a conservative bound on future vote movement. The rule separates two sources of uncertainty. It learns the trace-level switch probabilities that determine how much of the current margin is likely to be retained, while handling the harder question of where switching traces land through an adversarial bound calibrated from warmup traces. With true switch probabilities, MARS guarantees with high probability that the early-stopped answer matches the full-budget vote. In practice, a five-feature logistic model closely matches oracle switching behavior. Across three reasoning models and three competition-math benchmarks, MARS saves 25-47% of self-consistency tokens and 14-29% on top of DeepConf Online, a strong confidence-weighted baseline that already filters and truncates weak traces, while matching the accuracy of the corresponding full-budget baselines.

Spiking neural networks (SNNs) are promising for energy-efficient inference, and time-to-first-spike (TTFS) coding is especially attractive because each neuron fires at most once. In practice, however, this benefit is often reduced by the cost of computing a temporal decay term and multiplying it by the synaptic weight. We address this issue by turning a physical hardware "bug," the natural signal decay in optoelectronic devices, into the main computation of TTFS, named Otters++. Specifically, we use the measured decay of a custom In$_2$O$_3$ optoelectronic synapse to directly realize the TTFS temporal term, removing the need for explicit digital decay computation. To scale this idea to Transformer models, we establish a layer-wise functional equivalence between the Otters++ and a quantized neural network (QNN), and develop a hybrid training method that uses device-faithful SNN computation in the forward pass and QNN straight-through gradients through the equivalent QNN path in the backward pass, together with model distillation. This avoids differentiation through discrete first-spike events and reduces the over-sparsity problem in direct TTFS-SNN training. We further make training aware of measured device noise by sampling run-to-run variation, and refine the system-level energy model by accounting for device sharing and multi-hop communication. On GLUE dataset, Otters++ improves the average score to 84.17\% while maintaining a clear energy advantage over prior spiking Transformer baselines. These results show that physically grounded TTFS computing can be efficient, trainable, and robust under realistic hardware effects.

Large language model (LLM) routing has emerged as an effective paradigm for leveraging the complementary strengths of multiple LLMs through dynamic model and reasoning-strategy selection. Recent reinforcement learning (RL)-based routing methods further improve routing quality by optimizing routing policies from interaction feedback. However, they still struggle to provide informative and comparable learning signals under heterogeneous tasks with varying difficulty. In practice, multiple objectives (e.g., correctness, format behavior) are aggregated into a single scalar reward, leading to ambiguous credit assignment and conflicting optimization signals. Moreover, reward signals exhibit significant variability across instances, where some instances produce higher or more variable rewards, introducing optimization bias that favors trivial samples over informative ones. To address these issues, we propose \textbf{ReCal}, a \textbf{\underline{Re}}ward \textbf{\underline{Cal}}ibration framework for RL-based LLM routing. We first introduce a hierarchical reward decomposition mechanism with component-wise advantage estimation. We further propose a distribution-aware optimization strategy that calibrates optimization variability through variance-aware reweighting and per-dataset normalization. Experiments on seven datasets demonstrate that ReCal consistently improves routing performance, and training stability over baselines. Code is available at https://anonymous.4open.science/r/ReCal.

Large language models (LLMs) have demonstrated strong reasoning and instruction-following capabilities, making them potentially powerful tools for time-series analysis. However, time series lie outside their native textual modality, raising a fundamental question: how should time series be represented so that LLMs can reason about them effectively? Existing work typically serializes raw numerical sequences or fine-tunes pre-trained LLMs on time-series data. These approaches place the burden of extracting temporal structure directly on the LLM, creating a modality mismatch that often degrades performance on long sequences and introduces substantial computational overhead. In this work, we introduce Time-Series-to-Structured-Program representation (T2SP), a deterministic, training-free method that represents a time series as a structured symbolic program. T2SP decomposes time series into trends, periods, and salient events, expressing them in a program-friendly format aligned with the textual and code-like modalities on which LLMs are natively trained. By shifting temporal-structure extraction from the model to the representation itself, T2SP enables off-the-shelf LLMs to leverage their existing reasoning capabilities for time-series understanding. We evaluate T2SP on three reasoning tasks -- editing, captioning, and question answering -- where it consistently improves performance, reduces reasoning time, and lowers failure rates compared with raw-string representations. Our results demonstrate that T2SP provides an effective interface between time series and LLMs.

Offline preference optimization has become a practical substitute for reinforcement learning from human feedback, but pairwise objectives such as Direct Preference Optimization (DPO) and its variants use only the chosen and rejected responses stored in a static dataset. This leaves a useful signal unused: the response that the reference model itself would generate for the same prompt. We propose Direct Preference Optimization with Penalization (DPOP), a simple extension of DPO that augments the base preference loss with a gated penalty on reference-greedy responses. DPOP activates this penalty only when the current policy still assigns a lower likelihood to the preferred response than to the rejected response. On AlpacaEval 2.0, DPOP improves length-controlled win rate over DPO, SimPO, and AlphaDPO on both Llama-3-8b-it and Gemma-2-9b-it, achieving relative gains of 5.3\% and 4.4\% over baselines on the two models, respectively. Ablations further show that a SimNPO-style length-normalized penalty is stronger than NPO and token-level unlikelihood in this setting.

Auto-regressive models have emerged as powerful tools for sequential data, from language to video. Understanding how and why these models learn latent representations remains an open theoretical question. In this work, we demonstrate that when trained by empirical risk minimization on data from partially observed linear dynamical systems, two-layer linear auto-regressive models naturally learn to approximate Kalman filtering. In particular, we show that the learned hidden representation coincides, up to a similarity transformation, with the state estimates produced by the optimal (Kalman) filter, even though the model has no explicit knowledge of the underlying dynamics or state. The result follows from three main insights. First, we establish that the Kalman filter is well approximated by an auto-regressive model with bounded truncation error. Second, we show that despite non-convexity, the two-layer optimization landscape is benign, i.e., all stationary points are either strict saddles or global minima. Finally, as our main contributions, we provide finite-sample guarantees on prediction error, parameter estimation error, and latent state recovery. Numerical simulations support the theoretical results and demonstrate that the latent representations of auto-regressive models recover state estimates.

Masked diffusion language models (MDLMs) such as LLaDA now rival autoregressive (AR) LLMs, but every existing knowledge-editing and unlearning method (ROME, MEMIT, etc.) targets AR transformers and either makes assumptions that fail under iterative denoising, or requires gradient updates whose backward-pass activations cost tens of GB of extra VRAM and which collapse MDLMs at standard learning rates. We introduce TimeROME-DLM, the first training-free, gradient-free, inference-time knowledge-editing framework for MDLMs. It couples two components: a Temporal Indirect Effect (TIE) causal-tracing protocol that identifies, for each fact, the coordinate whose intervention most strongly drives the object prediction at later denoising steps; and a closed-form, low-rank residual edit memory that aggregates subject keys and target deltas across all forget facts and applies a single ridge-regularised update at that coordinate at every diffusion forward, with sparsification to limit utility spillover. Backbone weights stay frozen; only three hyperparameters (alpha, lambda, q) are tuned on a small validation split. On TOFU forget01 with TOFU-finetuned LLaDA-8B-Base, TimeROME-DLM cuts forget-set log-probability by roughly 83 nats. The same configuration transfers to LLaDA-8B-Instruct, Dream-7B, MMaDA-8B, DiffuLLaMA-7B, and LLaDA-MoE-1.4B. It keeps retain-set log-probability nearly flat (within ~1 nat at the utility-safe operating point) across 50 sequentially inserted facts, delivers a four- to fourteen-fold wall-clock speedup with zero additional VRAM over the strongest converged training-time baseline, and scales sub-linearly to 400 facts. TimeROME-DLM closes the locate-then-edit gap between AR LLMs and MDLMs at a fraction of the computational cost.

Low-Rank Adaptation (LoRA) significantly reduces compute and memory costs for finetuning Deep Learning models but is often harder to tune than dense training: when using factor-wise optimizers such as AdamW, it is sensitive to initialization choices, its optimal learning rates transfer poorly across ranks, and it often fails to beat dense baselines. We derive LoRA-Muon by applying the Muon optimizer's spectral steepest-descent rule to the low-rank setting. Along with our split weight-decay rule, our main claim is that LoRA-Muon is a good low-rank proxy for full-rank Muon and Shampoo-family optimizers. Its optimal learning rates transfer across rank, width, depth, and factor-rescaling. In our compute-matched TinyShakespeare study, a rank-$2$ proxy recovers the dense best tested learning rate, and a rank-$32$ LoRA-Muon run attains lower mean validation loss than the dense baseline in the seed-averaged sweep. We further show that the Spectron optimizer depends on arbitrary factor scaling, so it would likely be a poor fit when finetuning starts from badly imbalanced factors, and that LoRA-RITE's simplified QR-coordinate core implements the same spectral update. LoRA-Muon computes that update without QR-decomposition and avoids storing second moments, making it more accelerator-friendly and memory-efficient.

Granger Causal Discovery (GCD) is fundamental for analyzing temporal dependencies in complex systems. However, existing neural GCD methods predominantly rely on a "one-size-fits-all" paradigm, struggling to capture distribution shifts and dynamic regime changes inherent in real-world time series. This often leads to entangled representations and spurious causal graphs. In this paper, we propose CausalMoE, a billion-scale multimodal Granger causal foundation model that explicitly models patch-level heterogeneity. CausalMoE introduces a Pattern-Routed Mixture of Heterogeneous Experts, which dynamically identifies latent temporal patterns and routes patches to specialized domain experts, effectively decoupling regime-specific mechanisms from shared dynamics. To ensure interpretable graph recovery, we design a Causality-Aware Self-Attention mechanism operating across variables, yielding sparse Granger causal graphs via proximal optimization. Furthermore, CausalMoE is the first to integrate LLMs and VLMs to align numerical signals with textual and visual priors, regularizing causal estimation in complex scenarios. Extensive experiments demonstrate that CausalMoE establishes a new state-of-the-art on fully supervised benchmarks, while effectively generalizing to few-shot settings where traditional methods fail.

Emotion significantly influences cognition, enhancing memory and learning under certain conditions. Drawing on this principle, emotion-augmented deep learning investigates how affective states can improve neural network architectures and learning paradigms, achieving better generalization than non-emotional models. However, existing methods often rely solely on objective neurophysiological factors, neglecting the role of subjectivity in emotion. To bridge this gap, the present study introduces Emotional Regulation, a novel framework for modeling emotion in deep learning through artificial subjective experience. The method employs pre-training based on affective stimuli, balancing non-emotional and emotionally-influenced responses in downstream task optimization. Extensive experimentation was conducted in image classification, pre-training ResNet and ViT architectures on four emotional datasets, using CIFAR-10 and -100 as target benchmarks. Results reveal improvements over the aforementioned backbones, providing evidence of Emotional Regulation as a promising method for defining emotion-augmented deep learning through artificial subjective experience. Furthermore, the proposed approach overcomes the related work in image classification based on CIFAR, revealing Emotional Regulation as the new state-of-the-art in emotion-augmented deep learning for large-scale vision datasets. The study also enforces evidence of the impact of affective states in improving machine learning tasks' optimization, encouraging further investigation on emotion-inspired architectures.

Reinforcement learning has rapidly emerged as a key component in the training of reasoning and coding models, yet it remains poorly understood from a mechanistic perspective. We study how and through what underlying processes capabilities are acquired or enhanced via reinforcement learning post-training. Our analysis, based on controlled math reasoning experiments with Qwen-2.5-1.5B, reveals two core mechanisms: strategy selection and strategy improvement. Our results highlight the role of SFT data and reinforcement learning data in activating these mechanisms, in particular showing how supervising the model on diverse reasoning strategies can enable strategy selection and how increasing difficulty in reinforcement learning data can enable strategy improvement. Taken together, our results provide mechanistic insight into RL training and suggest practical interventions to continue scaling reasoning capabilities.

Large reasoning models (LRMs) improve complex problem-solving by generating long intermediate reasoning traces, but this substantially increases inference costs. NVFP4 inference offers a promising approach to reduce both computational and memory costs through hardware-supported low-precision execution. However, directly applying NVFP4 to LRMs introduces two practical limitations: reasoning accuracy degrades under quantization, and existing NVFP4 kernels do not fully realize latency benefits in small-batch autoregressive decoding. In this work, we analyze the effect of NVFP4 quantization on token-level uncertainty during reasoning. We show that quantization increases incorrect sampling at low-entropy symbolic tokens, while causing over-concentration on a small set of tokens in high-uncertainty reasoning steps. Based on this observation, we propose \textbf{ReSET}, a reasoning-step entropy-based temperature-scaling method that estimates step-level uncertainty online and adapts the decoding temperature using both token-level and step-level entropy signals. To address the latency gap, we further design a CUDA-core small-$M$ NVFP4 kernel for latency-critical autoregressive decoding. Across reasoning benchmarks and model scales, ReSET improves NVFP4 reasoning accuracy by up to $\sim\!$2 points over the NVFP4 baseline. Our CUDA-core small-$M$ kernel further improves latency-critical decoding, delivering up to $2.5\!\times$ kernel-level speedup over NVFP4 vLLM and approximately $2\!\times$ end-to-end decoding speedup over BF16. Code is available at https://github.com/aiha-lab/ReSET.

A key strength of diffusion models lies in their flexibility, since their outputs can be controlled at sampling time through guidance. However, beyond simple cases such as conditional sampling, the target distribution is often left implicit, defined only through a sampling rule or a heuristic energy function. To address this, we propose Jeffrey guidance, a principled framework that extends diffusion-model control to applications beyond what standard guidance can express. It leverages Jeffrey's rule of conditioning to update marginal distributions towards a prescribed target, preserving the conditional structure and minimally perturbing the joint distribution. We first demonstrate Jeffrey guidance by targeting a prescribed embedding distribution. With Inception embeddings as the target, this leads to substantial reductions in FID on both CIFAR-10 and FFHQ. We further apply Jeffrey guidance to fairness on CelebA-HQ, updating an unconditional diffusion model to enforce independence between attributes.

Weight-space geometry plays a central role in neural network optimization, yet manifold constraints are often applied uniformly across all weight matrices. In this work, we ask whether different transformer modules prefer different manifold geometries. We study Manifold Muon for GPT-2 pretraining and compare layer-wise assignments of Stiefel and DGram constraints across attention and MLP blocks. Our results show a clear asymmetry: constraining attention layers with Stiefel geometry while assigning DGram geometry to MLP layers gives the best performance among the tested configurations, whereas the inverted assignment and all-DGram configuration become unstable under the shared hyperparameter setting. We trace this failure to singular value growth in DGram-constrained attention weights, which can amplify attention logits and induce softmax saturation. These findings suggest that symmetry-aware and geometry-aware optimization for transformers should be module-specific rather than uniform.

We introduce Equilibrium State Estimation (ESE), a novel paradigm for simultaneous prediction, where multiple interacting systems require separate yet coordinated forecasts. Such scenarios often arise in real-world settings such as economics and healthcare modeling. Unlike existing approaches that predict one system at a time, ESE forecasts all systems in a single pass. It first estimates the equilibrium state across systems, then generates holistic forecasts based on the difference between the current state and the estimated equilibrium. Extensive experiments on synthetic and real-world datasets, including currency exchange and COVID-19 spread modeling, demonstrate that ESE is at least as accurate as state-of-the-art (SOTA) methods while being significantly faster. In addition, ESE integrates seamlessly with conventional predictors, combining their accuracy with its exceptional efficiency and delivering a 10-70x speedup. With linear-time complexity, ESE scales far better than SOTA methods as the number of systems increases. Moreover, it remains accurate under diverse perturbations, establishing ESE as a fast, generalizable, robust, and scalable multi-prediction method.

Contextual anomaly detection aims to identify abnormal behavior conditional on context variables, but practical deployments often face highly imbalanced context distributions where rare regimes can be critical information. Under such frequency bias, context-conditioned models can produce unstable decisions and excessive false alarms in rare contexts. We propose Rarity-Gated Feature-wise Linear Modulation (RGFiLM), a rarity-aware conditioning module that combines feature-wise modulation (i.e., context-conditioned scaling and shifting of hidden features) with a gate controlled by a data-driven rarity score. The rarity score is estimated from the empirical distribution of context variables and regulates how strongly context modulates intermediate representations: the gate becomes more decisive under rare contexts while remaining conservative under frequent contexts. We evaluate RGFiLM on maritime trajectory anomaly detection using AIS motion sequences with ERA5 environmental context in an environment-sensitive detour scenario. When instantiated in a sequential anomaly scoring pipeline, RGFiLM achieves the best mean F1--False Positive Rate (FPR) trade-off among the compared context-agnostic and context-conditioned methods. These results suggest that explicitly accounting for context rarity is an effective approach for reducing false alarms in context-sensitive anomaly detection.

While flow-based generative models have demonstrated strong performance across a wide range of domains, deploying them in safety-critical physical systems remains challenging due to strict constraint requirements. Existing approaches typically enforce safety through post-hoc corrections, which incur substantial computational overhead and may distort the learned distribution. We propose PolyFlow, a polytope-constrained flow matching framework that embeds constraints directly into the model and flow dynamics. PolyFlow introduces a discrete-time flow formulation and a projection-free architecture, which eliminate the discretization error and guarantee strict satisfaction of arbitrary polyhedral constraints, without the need for expensive iterative solvers. Experimental results show that PolyFlow achieves zero constraint violation while maintaining high distributional fidelity across a range of planning and control tasks. Compared to state-of-the-art constrained generation baselines, PolyFlow significantly reduces inference latency and demonstrates a favorable trade-off between safety, efficiency, and generative quality. Code is available on https://github.com/MJianM/PolyFlow.

We present MaxProof, a population-level test-time scaling framework for competition-level mathematical proof in the MiniMax-M3 series. M3 first trains three proof-oriented capabilities -- proof generation, proof verification, and critique-conditioned proof repair -- using a defense-in-depth generative verifier engineered for low false-positive rate. These capabilities are merged into a single released M3 model. At test time, MaxProof treats the model as a generator, verifier, refiner, and ranker, searches over a population of candidate proofs, and returns one final proof through tournament selection. With MaxProof test-time scaling, the M3 model reaches 35/42 on IMO 2025 and 36/42 on USAMO 2026, exceeding the human gold-medal threshold on both.

Anomaly detection in multivariate time series is challenged by four structurally distinct anomaly types -- point (isolated spikes), distributional (level shifts), temporal (rhythm changes), and collective (inter-sensor correlation breakdowns) -- each requiring different feature representations. Most unsupervised methods target only one or two types and provide limited interpretability. We present CRAFTIIF (Cross-Resolution Analytic Four-Type Interpretable Isolation Forest), a fully unsupervised framework targeting all four types without dataset-specific tuning. CRAFTIIF generates K=500 random analytic wavelet feature draws across four families (Morlet, DOG, Haar, Coiflet), each targeting a specific anomaly type, feeding five structured Isolation Forests -- one per type plus a meta-IF for compound anomalies. An adaptive Otsu/MAD threshold calibrates detection automatically across anomaly rates from 0.1% to 69.2%. Because each IF is trained exclusively on type-specific features, branch firing provides direct anomaly-type attribution by construction, without post-hoc explanation. Evaluated on all 19 datasets of the mTSBench benchmark (Zhou et al., TMLR 2026), CRAFTIIF achieves mean F1=0.228 (all 19 datasets) and F1=0.322 (13 detectable datasets), ranking first among all 25 evaluated methods on VUS-PR (0.463 vs. previous best 0.329, +40.7%). A diagnostic framework -- oracle F1, detectability limits, and branch separation ratios -- identifies 6 of 19 datasets as fundamentally undetectable by any unsupervised method. Ablation over 11 conditions confirms adaptive thresholding (+38% F1), four-branch structure (+20%), and meta-IF (+23%) are each essential. Code: https://github.com/smitswil/craftiif

Real-world time series are often highly incomplete and irregular due to sensor dormancy, transmission delays, and event-driven sampling, making reliable forecasting fundamentally challenging. Existing methods have evolved from impute-then-forecast pipelines to continuous-time models such as Neural ODEs and continuous-time graph networks. While these approaches improve the modeling of historical irregularity, they still rely on an implicit oracle assumption at inference time: the timestamps of future valid observations are presumed to be known in advance. This assumption limits practical relevance, since in many real systems the more fundamental question is not only what the future value will be, but also whether a valid observation will occur at all. In this paper, we propose Timeflies, a unified framework that reformulates forecasting as a joint problem of future observability inference and value estimation. To explicitly model the interaction between observation dynamics and state evolution, Timeflies adopts an observation stream and a value stream, coupled through three dedicated modules for reliability-aware embedding, observation-guided dependency modeling, and joint prediction. We further construct Shadow, a benchmark that combines natural missingness from public datasets with real-world industrial data, and introduce the Observation-Value Joint Entropy (OVJE) metric to comprehensively evaluate this coupled predictability. Extensive experiments show that Timeflies consistently outperforms existing methods, highlighting the importance of explicitly modeling future observability in time series forecasting with missing values. Code and dataset are available in https://github.com/ant-intl/Timeflies.

Chain-of-thought (CoT) reasoning is the dominant paradigm for inference-time scaling in language models, yet the causal influence of individual steps on the final answer poorly understood. We estimate each step's causal importance via early exit and use this measure to study how answers form across the reasoning traces of several model families. Across diverse tasks, we find that reasoning typically crosses a \emph{commitment boundary} -- a sharp transition from transient intermediate guesses to a stable, high-confidence answer. This transition often happens in a single step, well before the model's reasoning block ends, and is followed by \emph{epiphenomenal} CoT steps that leave the final answer probability unaltered. Using attention probes, we show that answer-formation stages can be linearly decoded from intermediate reasoning steps with high accuracy and generalize robustly to unseen reasoning tasks. We exploit this signal to early-exit reasoning blocks at the commitment boundary, reducing the length of CoTs up to 55\% on average with negligible impact on model performance.

Masking-based post-hoc explanation methods, such as KernelSHAP and LIME, estimate local feature importance by querying a black-box model under randomized perturbations. This paper formulates this procedure as communication over a query channel, where the latent explanation acts as a message and each masked evaluation is a channel use. Within this framework, the complexity of the explanation is captured by the entropy of the hypothesis class, while the query interface supplies information at a rate determined by an identification capacity per query. We derive a strong converse showing that, if the explanation rate exceeds this capacity, the probability of exact recovery necessarily converges to one in error for any sequence of explainers and decoders. We also prove an achievability result establishing that a sparse maximum-likelihood decoder attains reliable recovery when the rate lies below capacity. A Monte Carlo estimator of mutual information yields a non-asymptotic query benchmark that we use to compare optimal decoding with Lasso- and OLS-based procedures that mirror LIME and KernelSHAP. Experiments reveal a range of query budgets where information theory permits reliable explanations but standard convex surrogates still fail. Finally, we interpret super-pixel resolution and tokenization for neural language models as a source-coding choice that sets the entropy of the explanation and show how Gaussian noise and nonlinear curvature degrade the query channel, induce waterfall and error-floor behavior, and render high-resolution explanations unattainable.

The advancement of Large Reasoning Models (LRMs) has catalyzed a paradigm shift from reactive ``fast thinking'' text generation to systematic, step-by-step ``slow thinking'' reasoning, unlocking state-of-the-art performance in complex mathematical and logical tasks. However, the field faces \textit{the fundamental gap between token-level behavioral analysis and internal reasoning mechanisms, and the instability of reinforcement learning (RL) for reasoning optimization relying on costly external verifiers}. We identify and formally define \textbf{Entropy-Gradient Inversion}, a robust negative correlation between token entropy and logit gradients that acts as a definitive geometric fingerprint for LRM reasoning capability. Building on this, we propose \textbf{Correlation-Regularized Group Policy Optimization (CorR-PO)}, which embeds this inversion signature into RL reward regularization. Extensive experiments on various reasoning benchmarks across multiple model scales show CorR-PO consistently outperforms state-of-the-art baselines, confirming that stronger inversion directly correlates with superior reasoning performance.

Majority voting over sampled answers is the dominant unsupervised aggregator for multi-sample LLM inference. In this paper, we show a delegation-based aggregator (Propagational Proxy Voting, PPV; Sakai et al., 2025) yields an unsupervised consensus rule that beats majority on MMLU-Pro by +1.5 pp overall and +2.24 pp on the non-trivial subset (paired McNemar p ~ 1.0e-14, n = 8,099). Majority discards two signals that every sample carries: within-group letter entropy and between-group reasoning geometry. PPV exposes per-voter levers that consume exactly these two signals: When (how much weight a voter keeps on its own pick) and Whom (how it splits the remainder across peers). We drive When with letter entropy and Whom with per-question-centered embedding cosine. Our method needs no gold labels and no auxiliary training: per-question, we partition 128 sampled generations into 16 groups, compute each group's letter-level semantic entropy and reasoning embedding centroid, and feed both into a stochastic delegation matrix whose stationary distribution selects the consensus answer. We walk through an example in which PPV overturns a clear 10-6 majority for the wrong letter: the 10-voter majority cluster is geometrically incoherent (mean within-cluster cosine -0.02) while the 6-voter minority is tight (+0.26), so propagated delegation mass concentrates on the minority's answer even though entropy alone would keep the majority ahead. We further report delegation strategies with negative results that constrain the design space for unsupervised LLM aggregation. No within-question ensemble of confidence modes closes the oracle gap.

Unlike conventional "black-box" transformers with classical self-attention mechanism, we build a lightweight and interpretable transformer-like neural net by unrolling a mixed-graph-based optimization algorithm to forecast traffic with spatial and temporal dimensions. We construct two graphs: an undirected graph $\mathcal{G}^u$ capturing spatial correlations across geography, and a directed graph $\mathcal{G}^d$ capturing sequential relationships over time. We predict future samples of signal $\mathbf{x}$, assuming it is "smooth" with respect to both $\mathcal{G}^u$ and $\mathcal{G}^d$, where we design new $\ell_2$ and $\ell_1$-norm variational terms to quantify and promote signal smoothness (low-frequency reconstruction) on a directed graph. We design an iterative algorithm based on alternating direction method of multipliers (ADMM), and unroll it into a feed-forward network for data-driven parameter learning. We periodically insert graph learning modules for $\mathcal{G}^u$ and $\mathcal{G}^d$ that play the role of self-attention. Experiments show that our unrolled networks achieve competitive traffic forecast performance as state-of-the-art prediction schemes, while reducing parameter counts drastically.

Learning effective representations of urban environments requires capturing spatial structure beyond fixed administrative boundaries. Existing geospatial representation learning approaches typically aggregate Points of Interest (POIs) into pre-defined administrative regions such as census units or ZIP code areas, assigning a single embedding to each region. However, POIs often form semantically meaningful groups that extend across, within, or beyond these boundaries, defining places that better reflect human activity and urban function. To address this limitation, we propose PlaceRep, a geospatial representation learning method that constructs place-level representations by clustering spatially and semantically related POIs. PlaceRep summarizes large-scale POI graphs from U.S. Foursquare data to produce general-purpose urban region embeddings while automatically identifying places across multiple spatial scales. By eliminating model pre-training, PlaceRep provides a scalable and efficient solution for multi-granular geospatial analysis. Experiments using the tasks of population density estimation and housing price prediction as downstream tasks show that PlaceRep outperforms most state-of-the-art graph-based geospatial representation learning methods and achieves up to a x100 speedup in generating region-level representations on large-scale POI graphs. The implementation of PlaceRep is available at https://github.com/mohammadhashemii/PlaceRep.

In-context learning enables transformer models to generalize to new tasks based solely on input prompts, without any need for weight updates. However, existing training paradigms typically rely on large, unstructured datasets that are costly to store, difficult to evaluate for quality and balance, and pose privacy and ethical concerns due to the inclusion of sensitive information. Motivated by these limitations and risks, we propose an alternative training strategy where we leverage a collection of multiple, small-scale, and domain-specific datasets. We empirically demonstrate that the increased quality and diversity of such data improve the generalization abilities of in-context learners beyond their training domain, while achieving comparable performance with models trained on a single large-scale dataset. We investigate this paradigm by leveraging meta-learning to train an in-context learner on the Meta-Album collection under several settings. Firstly, we show the performance in a controlled environment, where the test domain is completely excluded from the training knowledge. Secondly, we explore the robustness of these models to forgetting in a continual scenario where the information is accessible for a limited time. Finally, we explore the more challenging unsupervised scenario. Our findings demonstrate that transformers still generalize for in-context prediction when trained on a curated dataset collection while offering advantages in modularity and replaceability.

Bifurcation phenomena in nonlinear dynamical systems often lead to multiple coexisting stable solutions, particularly in the presence of symmetry breaking. Deterministic machine learning models are unable to capture this multiplicity, averaging over solutions and failing to represent lower-symmetry outcomes. In this work, we formalize the use of generative AI, specifically flow matching, as a principled way to model the full probability distribution over bifurcation outcomes. Our approach builds on existing techniques by combining flow matching with equivariant architectures and an optimal-transport-based coupling mechanism. We generalize equivariant flow matching to a symmetric coupling strategy that aligns predicted and target outputs under group actions, allowing accurate learning in equivariant settings. We validate our approach on a range of systems, from simple conceptual systems to physical problems such as buckling beams and the Allen--Cahn equation. The results demonstrate that the approach accurately captures multimodal distributions and symmetry-breaking bifurcations. Moreover, our results demonstrate that flow matching significantly outperforms non-probabilistic and variational methods. This offers a principled and scalable solution for modeling multistability in high-dimensional systems.

Diffusion LLMs (dLLMs) have recently emerged as a powerful alternative to autoregressive LLMs (AR-LLMs) with the potential to operate at significantly higher token-generation rates. To unlock this potential, we present Spiffy, a speculative decoding algorithm to accelerate dLLM inference while provably preserving the model's output distribution. This work addresses the unique challenges involved in applying ideas from speculative decoding of AR-LLMs to dLLMs. Spiffy performs auto-speculation to eliminate the overheads of an independent draft model, structuring draft states in the form of a novel directed draft graph to take advantage of the bidirectional, blockwise nature of dLLM generation. These draft graphs are calibrated offline to maximize acceptance rates and are dynamically pruned during inference for improved computational efficiency. We present a detailed formulation of Spiffy and demonstrate its ability to accelerate LLaDA, Dream, and SDAR models in combination with KV caching and threshold-based dynamic unmasking leading to up to $8.6\times$ reduction in model inferences and $6.3\times$ acceleration in token rate.

Proteins inherently possess a consistent sequence-structure duality. The abundance of protein sequence data, which can be readily represented as discrete tokens, has driven fruitful developments in protein language models (pLMs). A key remaining challenge, however, is how to effectively integrate continuous structural knowledge into pLMs. Current methods often discretize protein structures to accommodate the language modeling framework, which inevitably results in the loss of fine-grained information and limits the performance potential of multimodal pLMs. In this paper, we argue that such concerns can be circumvented: a sequence-based pLM can be extended to incorporate the structure modality through continuous tokens, i.e., high-fidelity protein structure latents that avoid vector quantization. Specifically, we propose a hybrid diffusion protein language model, HD-Prot, which embeds a continuous-valued diffusion head atop a discrete pLM, enabling seamless operation with both discrete and continuous tokens for joint sequence-structure modeling. It captures inter-token dependencies across modalities through a unified absorbing diffusion process, and estimates per-token distributions via categorical prediction for sequences and continuous diffusion for structures. Extensive results demonstrate that HD-Prot achieves competitive performance in unconditional sequence-structure co-generation, motif-scaffolding, protein structure prediction, and inverse folding tasks. Furthermore, our method can perform on par with state-of-the-art multimodal pLMs, despite being developed under limited computational resources (i.e., less than one-tenth the budget for modality extension fine-tuning). It highlights the viability of simultaneously estimating categorical and continuous distributions within a unified language model architecture, offering a promising alternative direction for multimodal pLMs.

Synthetic appliance data are essential for developing non-intrusive load monitoring algorithms and enabling privacy preserving energy research, yet the scarcity of labeled datasets remains a significant barrier. Recent GAN-based methods have demonstrated the feasibility of synthesizing load patterns, but most existing approaches treat all devices uniformly within a single model, neglecting the behavioral differences between intermittent and continuous appliances and resulting in unstable training and limited output fidelity. To address these limitations, we propose the Cluster Aggregated GAN framework, a hybrid generative approach that routes each appliance to a specialized branch based on its behavioral characteristics. For intermittent appliances, a clustering module groups similar activation patterns and allocates dedicated generators for each cluster, ensuring that both common and rare operational modes receive adequate modeling capacity. Continuous appliances follow a separate branch that employs an LSTM-based generator to capture gradual temporal evolution while maintaining training stability through sequence compression. Extensive experiments on the UVIC smart plug dataset demonstrate that the proposed framework consistently outperforms baseline methods across metrics measuring realism, diversity, and training stability, and that integrating clustering as an active generative component substantially improves both interpretability and scalability. These findings establish the proposed framework as an effective approach for synthetic load generation in non-intrusive load monitoring research.

The decentralization of autoregressive generation has attracted considerable attention in recent years as a solution to scaling bottlenecks. However, despite promising empirical results, this paradigm currently lacks rigorous theoretical justification. In this work, we formally establish the theoretical equivalence between decentralized and centralized training. To achieve this, we adapt the Discrete Flow Matching framework for autoregressive generation, leveraging its inherent properties to demonstrate that global models naturally decompose into independent experts. Finally, we conduct extensive experiments across diverse multimodal benchmarks, empirically validating that decentralized training maintains competitive parity with standard centralized architectures.

Multimodal variational autoencoders (VAEs) are widely used for weakly supervised generative learning with multiple modalities. Predominant methods aggregate unimodal inference distributions using either a product of experts (PoE), a mixture of experts (MoE), or their combinations to approximate the joint posterior. In this work, we revisit multimodal inference through the lens of probabilistic opinion pooling, an optimization-based approach. We start from Hölder pooling with $α=0.5$, which corresponds to the unique symmetric member of the $α\text{-divergence}$ family, and derive a moment-matching approximation, termed Hellinger. We then leverage such an approximation to propose HELVAE, a multimodal VAE that avoids sub-sampling, yielding an efficient yet effective model that: (i) learns more expressive latent representations as additional modalities are observed; and (ii) empirically achieves better trade-offs between generative coherence and quality, outperforming state-of-the-art multimodal VAE models.

The high-level concepts that a neural network uses to perform computation need not be aligned to individual neurons (Smolensky, 1986). Language model interpretability research has thus turned to techniques which decompose the neuron basis into more interpretable units of model computation, such as sparse autoencoders (SAEs). However, not all neuron-based representations are uninterpretable. For the first time, we empirically show that MLP neurons are as sparse a feature basis as SAEs. We use this finding to develop an end-to-end gradient-based attribution pipeline for circuit tracing on the MLP neuron basis, which surfaces causally effective neurons on a variety of tasks. On a standard subject-verb agreement benchmark (Marks et al., 2025), a circuit of $\approx 10^2$ MLP neurons is enough to control model behaviour. On the multi-hop city-state-capital task from (Lindsey et al., 2025), we find a circuit in which small sets of neurons encode specific latent reasoning steps (e.g. mapping a city to its state), and can be steered to change the model's output. This work thus advances automated interpretability of language models without imposing additional training costs.

The Strong Lottery Ticket Hypothesis (SLTH) states that large, randomly initialized neural networks contain sparse subnetworks capable of approximating a target function at initialization without training, suggesting that pruning alone is sufficient. Pruning methods are typically classified as unstructured, where individual weights can be removed from the network, and structured, where parameters are removed according to specific patterns, as in neuron pruning. Existing theoretical results supporting the SLTH rely almost exclusively on unstructured pruning, showing that logarithmic overparameterization suffices to approximate simple target networks. In contrast, neuron pruning has received limited theoretical attention, despite its practical appeal for direct hardware speedups. In this work, we consider the problem of approximating a single bias-free ReLU neuron by pruning hidden units of a randomly initialized two-layer ReLU network, effectively isolating the intrinsic limitations of neuron pruning. We show that achieving an $\varepsilon$-approximation requires a starting network size of $Ω(1/\varepsilon)$ for neuron pruning, whereas weight pruning succeeds with only $O(\log(1/\varepsilon))$ hidden units, revealing an exponential separation between the two approaches.

Time-series anomaly detection (TSAD) is critical in domains such as industrial monitoring, healthcare, and cybersecurity, but it remains challenging due to rare and heterogeneous anomalies and the scarcity of labelled data. This scarcity makes unsupervised approaches predominant, yet existing methods often rely on reconstruction or forecasting, which struggle with complex data, or on embedding-based approaches that require domain-specific anomaly synthesis and fixed distance metrics. We propose ASTER, a framework that generates pseudo-anomalies directly in the latent space, avoiding handcrafted anomaly injections and the need for domain expertise. A latent-space decoder produces tailored pseudo-anomalies to train a Transformer-based anomaly classifier, while a pre-trained LLM enriches the temporal and contextual representations of this space. Experiments on three benchmark datasets show that ASTER achieves state-of-the-art performance and sets a new standard for LLM-based TSAD.

Brain MRI underpins a wide range of neuroscientific and clinical applications, yet most learning-based methods remain task-specific and require substantial labeled data. Here we show that a single self-supervised representation can generalize across heterogeneous brain MRI endpoints. We trained BrainDINO, a self-distilled foundation model, on approximately 6.6 million unlabeled axial slices from 20 datasets encompassing broad variation in population, disease, and acquisition setting. Using a frozen encoder with lightweight task heads, BrainDINO supported transfer across tumor segmentation, neurodegenerative and neurodevelopmental conditions classification, brain age estimation, post-stroke temporal prediction, molecular status prediction, MRI sequence classification, and survival modeling. Across tasks and supervision regimes, BrainDINO consistently equaled or exceeded natural-image and MRI-specific self-supervised baselines, with particularly strong advantages under label scarcity. Representation analyses further showed anatomically organized and pathology-sensitive feature structure in the absence of task-specific supervision. Our findings indicate that large-scale slice-wise self-supervised learning can yield a unified brain MRI representation that supports diverse neuroimaging tasks without volumetric pretraining or full-network fine-tuning, establishing a scalable foundation for robust and data-efficient brain imaging analysis. Code is available at https://github.com/mclwu22/BrainDINO

Deep neural networks achieve impressive results across diverse applications, yet their overconfidence on unseen inputs necessitates reliable epistemic uncertainty modeling. Existing methods for uncertainty modeling face a fundamental dilemma: Bayesian approaches provide principled estimates but remain computationally prohibitive, while efficient second-order predictors lack rigorous connections between their specific objectives and epistemic uncertainty quantification. To resolve this dilemma, we introduce Dirichlet-approximated possibilistic posterior predictions (DAPPr), a principled framework grounded in possibility theory. We define a possibilistic posterior over parameters, project it to the prediction space via supremum operators, and approximate the projected posterior using learnable Dirichlet possibility functions. This projection-and-approximation strategy yields a simple training objective with closed-form solutions. Despite its simplicity, extensive experiments across diverse benchmarks show that DAPPr achieves competitive or superior uncertainty quantification performance over state-of-the-art second-order predictors while maintaining both principled derivation and computational efficiency. Code is available at https://github.com/MaxwellYaoNi/DAPPr.

Scaling LLMs requires tremendous computational resources, and recent advances in AI have gone hand in hand with massive amounts of capital expenditure. While it is established that scaling up LLMs reliably increases model quality (quantified in terms of loss or downstream evaluations), it is unclear how these quality improvements translate to potential revenue, and whether revenue increases would offset costs of larger-scale training and inference. In this work, we develop an economic model for characterizing the rational behavior of an LLM training firm by combining scaling laws with microeconomic theory. Under our model of firm behavior, LLM quality can be increased with more parameters and training tokens, leading to more potential adoption by consumers, who each have a quality threshold for using the LLM. On the other hand, additional parameters and training tokens both incur additional costs. We analyze the profit maximization problem for this model under compute-bound and data-bound regimes. In the compute-bound regime, optimal model size and token budget track hardware efficiency $E$ (FLOPs/\$) at a near-linear rate; total training cost then scales sub-quadratically in $E$. Data efficiency improvements incentivize larger models and training expenditure. When we are limited to $D$ data, profit-optimal training expenditure scales as $D^2/E$, i.e, increase with data and decreases with hardware efficiency (as well as data efficiency). Finally, we analyze practical trends in training expenditure: current trends are consistent with our most permissive model variants in the compute-bound regime, but are not profit-optimal in the data-bound regime or assuming hardware advances will stall. Overall, our results provide a theory of profit-optimal LLM training, providing a foundation for engaging critically with industry statements and supporting long-term economic decision making.

Non-monotonic sequence generation methods, such as masked diffusion models, provide a flexible alternative to left-to-right autoregressive modeling by allowing tokens to be generated in non-fixed and prescribed orders. Despite their practical advantages, most existing non-monotonic models are order-agnostic and rely on a fixed-length grid, limiting their ability to support variable-length generation and adaptive insertion order. In this work, we introduce a probabilistic framework for learning insertion order in variable-length insertion models. We formalize a bijective correspondence between insertion trajectories and permutations, which enables an exact reparameterization of the data likelihood as a sum over permutations. Building on this result, we propose the Insertion Process (IP), a stochastic generative model that jointly learns where to insert, what to insert, and when to terminate, trained via permutation-based variational inference. Unlike prior fixed-canvas approaches, IP natively supports variable-length generation and learns data-driven preferences over insertion orders. Experiments on goal-conditioned planning and molecular string generation demonstrate that learning insertion order improves both modeling quality and generalization in domains without a canonical left-to-right structure.

This paper presents a novel data-free and training-free compression approach for speech foundation models using channelwise clustering via k-means. More fine-grained, mixed sparsity pruning by layer-level varying number of parameter clusters is also explored. Experiments conducted on the LibriSpeech dataset suggest that when operating with pruning sparsity of 50% on HuBERT-large, consistent WER reductions of 27.73%/18.61% absolute (34.37%/21.91% relative) over the magnitude-based pruning were obtained on the test-clean and test-other subsets before fine-tuning and 0.19%/0.79% absolute (3.36%/4.62% relative) after fine-tuning with only 3 epochs. Similar WER reductions of 2.86%/5.02% absolute (59.21%/55.29% relative) were observed against magnitudebased pruning on Whisper-large-v3 at 10% sparsity, all with no significant WER increase relative to the uncompressed baseline.

Generative listwise ranking with Large Multimodal Models (LMMs) aims to capture global list context in a single forward pass, but its effectiveness degrades in long-context multimodal scenarios. We identify a recurring failure mode, parse collapse, where the autoregressive decoder produces fluent yet incomplete rankings by silently omitting candidates and terminating early. This failure stems from limited context utilization rather than simple formatting mistakes, making prompt engineering and constrained decoding insufficient. We propose PRISMR (Parameterized Representation Internalization for Semantic Multimodal Ranking), a framework that replaces transient in-context list processing with parametric structural conditioning. PRISMR uses a lightweight hypernetwork to encode multimodal candidates in parallel and generate item-specific LoRA weights, which are synthesized into an instance-specific adapter for a LMM. This paradigm enables more robust internalization of list structure while preserving the base model. We further introduce a large-scale multimodal review-ranking benchmark for evaluation. Experiments demonstrate that PRISMR substantially reduces parse collapse, improves listwise ranking performance, and transfers effectively across domains and instruction-tuned backbones.

Reinforcement Learning with Verifiable Rewards (RLVR) is a central technique for improving long-horizon reasoning in Large Language Models (LLMs). However, existing RLVR methods often encourage unnecessarily long reasoning rollouts, which can degrade reasoning coherence and exhaust the available context budget. Existing approaches to long-context organization often depend on external mechanisms to organize rollouts, rather than enabling the model to manage its own reasoning trajectory. To address this limitation, we propose ReSum, a novel RLVR framework that enables LLMs to compress and organize their reasoning trajectories through self-summarization. Our pilot studies show that self-summarization stabilizes generation by lowering token-level entropy, and that introducing a ``summarization'' phrase can substantially mitigate errors propagated from an incorrect rollout prefix. Motivated by these findings, ReSum adopts a summarization-aware adaptive rollout mechanism that contrastively evaluates whether self-summarization benefits the ongoing reasoning process. Specifically, when the model spontaneously triggers self-summarization, ReSum masks the summarization phrase to create a contrastive branch; for non-summarization positions, it instead randomly injects the phrase to create a matched branch. We further design a summarization-aware advantage to enable finer-grained comparison between contrastive rollout trajectories. Extensive experiments show that ReSum improves performance at an average of 4\% while reducing rollout length by 18.6\%.

Retrieval augmented generation (RAG) depends critically on the quality and granularity of retrieved evidence. Large retrieval units preserve context but often introduce irrelevant content, which can dilute answer bearing evidence and worsen long context utilization. Fine-grained units are more compact, but they may be difficult to retrieve reliably because short chunks can lack semantic, lexical, or bridging cues needed to match the query. We propose Uncertainty-aware Multi-Granularity RAG (UMG-RAG), a training-free hybrid retrieval framework that treats chunk granularity as query-specific reliability estimation. Instead of training a new retriever or modifying the generator, UMG-RAG uses existing dense and sparse retrievers as complementary experts across multiple chunk granularities. For each query, it converts each expert-granularity score list into an evidence distribution, estimates reliability from distribution entropy, and fuses candidates according to query-specific semantic, lexical, and granularity confidence. We further introduce UMGP-RAG, a parent promotion variant that uses fine-grained hits to locate relevant evidence while returning broader non-redundant parent chunks for local coherence. Experiments on question answering benchmarks show that uncertainty-aware fusion and parent promotion improve generation quality while maintaining a lightweight, plug-and-play retrieval pipeline.

Large Vision-Language Models (LVLMs) have achieved strong performance across medical imaging tasks, yet they remain prone to factual inconsistencies, poor visual grounding, and misalignment with clinically meaningful feedback. Existing post-training alignment approaches, including Direct Preference Optimization (DPO) and its variants, face three critical limitations in the medical domain: (1) sequence-level reward signals treat clinically critical tokens identically to generic filler text; (2) reliance on static supervised fine-tuning references as preferred responses introduces an off-policy distribution shift, steering optimization toward stylistic artifacts over clinical correctness; and (3) alignment objectives lack explicit visual grounding constraints, leaving models insensitive to subtle yet diagnostically decisive pathological features. Our method leverages a bidirectional token-wise KL regularizer alongside a visual-contrastive grounding objective that pairs clean and lesion-corrupted images to penalize responses generated without adequate visual evidence. Together, these components form a fine-grained, on-policy alignment framework that constructs preference pairs by minimally editing model-generated outputs, correcting only clinically erroneous spans while preserving the original linguistic style. Extensive experiments across medical imaging tasks and clinical text generation benchmarks validate the effectiveness of our approach.

Are large language models (LLMs) bad at capturing human judgment? Two commonly stated limitations are that LLMs fail to capture full distributions of responses, and that their judgments are unstable across wording variations. We demonstrate simple prompting strategies that mitigate these limitations. Across two datasets--a U.S.-representative set of 144 moral scenarios and 38 moral beliefs from the International Social Survey Programme's Family and Changing Gender Roles module covering 32 countries--we show how simple elicitation techniques help improve AI-human alignment. First, prompting models to report standard deviations and response proportions recovers the full range of human responses better than common strategies. Second, ensuring scenarios are clear to human participants--as reflected in human confusion ratings--boosts model alignment, and LLMs can track human confusion ratings. At the same time, we find that LLMs' estimates of their own error are poorly calibrated, though they can predict human variability relatively well. These results suggest that asking better questions to LLMs can yield better answers.

Irrelevant numbers in a prompt can shift language model judgments, producing anchoring effects in numerical reasoning. We study where this anchor-sensitive signal is carried inside language models using a controlled multiple-choice setup with shared answer options. We define a logit-difference metric comparing the correct answer option with the answer option corresponding to the anchor, and validate that it tracks behavioral anchoring. Using attribution-based circuit localization on 7B--8B Qwen and Llama base and instruction-tuned models, we find that edge-level methods recover this signal more faithfully than node-level methods. Low- and high-anchor circuits transfer strongly within a model, suggesting shared pathway structure across anchor direction. However, sparse transfer across base and instruction-tuned variants is less reliable, indicating that post-training changes which pathways matter most. Overall, our results provide a mechanistic account of how anchoring-related decision signals are carried inside language models.

Moving instance segmentation (MIS) attracts increasing attention due to its broad applications in traffic surveillance, autonomous driving, and animal tracking. Event cameras record asynchronous brightness changes, providing high temporal resolution and dynamic range, which makes them highly sensitive to motion information. By fusing event and image features, motion cues from events can complement spatial details from images, enhancing the performance of MIS. However, current multimodal MIS methods still struggle to segment small moving instances, as event cameras often yield sparse features under limited resolution. Moreover, event features entangle appearance attributes with motion cues, which further restricts effective cross-modal fusion. To address these challenges, we first propose a dual-disentangling feature extraction framework that separates and extracts appearance and motion information within both image and event modalities, thereby improving feature density. Subsequently, a multi-granularity cross-modal alignment is introduced to align distributionally and semantically consistent features across modalities, enabling more effective fusion with rich spatial and temporal details. The experiment results demonstrate that our method achieves state-of-the-art performance in multimodal MIS, especially for small instances under challenging conditions such as fast motion and low-light settings.

Interleaved thinking, where a unified multimodal model alternates between textual reasoning and visual generation, has shown promise on spatial and physical tasks. However, in complex long-chain scenarios, we identify a fundamental failure mode: generated images diverge from the textual context while subsequent text ignores the visual evidence, causing the two modalities to alternate without genuinely informing each other. We term this Modal Isolation and attribute it to compounding information loss at modality boundaries. We decompose each reasoning cycle into atomic operations and define modality transition loss, quantifying cross-modal hallucination (text-to-image) and visual utilization deficit (image-to-text) at each boundary. We propose MoTiF (Modality Tiransition Fidelity), a two-stage training framework that directly optimizes these transitions: Reflective SFT trains the model to detect and recover from erroneous visual outputs; Flow-GRPO improves image generation fidelity via reinforcement learning. All training signals in MoTiF derive from transition-level fidelity rather than end-task accuracy. Across four visual puzzle benchmarks, this transition-level supervision substantially improves both cross-modal coherence and final task accuracy. The results demonstrate that effective interleaved reasoning requires explicit structural supervision at modality boundaries, not merely scaling or end-task optimization.

Autoregressive video diffusion models provide a natural formulation for streaming and variable-length video generation by conditioning newly generated frames on previously generated content. However, extending these models to minute-level generation remains challenging: the limited KV-cache budget prevents the model from retaining the full history, while repeatedly conditioning on self-generated frames induces a context distribution shift that accumulates over time, leading to visual artifacts, quality degradation, and temporal drift. In this paper, we propose TetherCache, a training-free and plug-and-play cache management strategy for drift-resistant long video generation. TetherCache organizes the cache into sink, memory, and recent regions, and introduces two complementary mechanisms. First, GRAB (Gated Recall with Attention-Diversity Balancing) selects long-range memory frames using a gated score that combines attention-based relevance with temporal diversity, preserving informative yet diverse historical context under a fixed cache budget. Second, TAME (Trusted Alignment via Memory Editing) lightly edits newly recalled memory tokens by aligning their statistics to a trusted context distribution, reducing the pollution caused by drifted historical features. Built on Self-Forcing, TetherCache consistently improves long-video generation quality on VBench-Long across 30s, 60s, and 240s settings. In particular, for 240s generation, it substantially improves overall and semantic scores while reducing quality drift from 7.84 to 1.33, demonstrating its effectiveness for stable long-horizon autoregressive video diffusion.

While Large Language Models (LLMs) have advanced open-domain dialogue systems, maintaining long-term consistency remains a challenge due to inherent limitations in long-context reasoning and the inefficiency of processing extensive raw text. Existing approaches typically rely on either unstructured memory storage, which is prone to information loss, or computationally expensive LLMs that incur high latency. To address these limitations, we propose G-Long, a graph-enhanced framework that utilizes a fine-tuned small Language Model (sLM) for structured triplet extraction and associative retrieval, significantly reducing operational costs. Furthermore, we introduce the novel attention-aware importance scoring mechanism that leverages the intrinsic cross-attention signals of a T5 summarizer to identify salient memories. Extensive experiments across diverse benchmarks demonstrate that G-Long achieves state-of-the-art performance in both response generation and memory retrieval, yielding performance gains of up to 9.8% in response quality on MSC and 40.8% in retrieval recall on LME, while significantly minimizing computational overhead.

Simultaneous speech-to-speech translation aims to enable near-real-time communication by minimizing latency, offering a compelling, real-time alternative to the high latency of consecutive translation. However, the excessive pursuit of low latency often results in fragmented chunk-wise speech. Consequently, listeners are subjected to an unnatural acoustic flow punctuated by frequent pauses, which could increase their cognitive load. To bridge this gap, we introduce a fluency-aware optimization framework designed to discover the sweet spot between the low-latency benefits of simultaneous translation and the natural flow of consecutive translation. Our framework minimizes inter-chunk silences by leveraging model-internal signals, including linguistic diversity and induced temporal variability in speech durations. Experiments on short- and long-form benchmarks show that our framework produces natural speech flow while maintaining competitive latency and translation quality.

Vision-language models (VLMs) achieve strong singleshot spatial grounding, yet lack any mechanism to observe and correct their own predictions. We find that naively prompting a VLM to iterate over rendered visualizations of its predictions causes catastrophic failure: Acc@0.5 on referring expression comprehension collapses from 79.6% to 48.7% (a 31 percentage point drop), revealing a fundamental gap between grounding capability and self-correction ability. We propose Iterative Visual Thinking (IVT), a closed-loop framework in which the model predicts a bounding box, observes the prediction rendered on the image, and iteratively refines through visual feedback. A two-phase training recipe closes the self-correction gap: first, we exploit the base model's own predictions as realistic errors and prompt a teacher VLM to generate corrective reasoning traces, yielding supervised data without human annotation; second, we apply Group Relative Policy Optimization (GRPO) with a simple IoU reward to stabilize multi-step refinement. On a mixed benchmark spanning RefCOCOg, Ref-Adv, and Ref-L4 (505 test samples), SFT warm-up with IVT surpasses the single-shot base model on every metric: Acc@0.5 rises to 82.0% (+2.4pp), Acc@0.7 to 74.1% (+3.2pp), and Acc@0.9 to 48.3% (+2.8pp). GRPO further reduces per-step IoU degradation by 5x, stabilizing the refinement trajectory. All training uses only 2,400 samples on a single GPU, demonstrating that spatial self-correction is a learnable capability that can be instilled at modest scale.

The rapid updates of online news make tracking event developments challenging, highlighting the need for timeline summarization (TLS). Hallucinations, where LLM-generated content deviates from source news, still remain a critical issue in LLM-based TLS and are not well studied in existing works. To bridge this gap, we identify two primary types of hallucinations: unfaithful content during news summarization and information omission in date-event summarization. Then, we propose NTS-CoT, a novel framework that leverages Chain-of-Thought (CoT) reasoning to mitigate hallucinations in TLS. The framework consists of three key modules: i) Element-CoT to capture essential news elements for faithful summarization, ii) Date Selection to combine temporal saliency and event prominence for timestamp selection, and iii) Causal-CoT to infer causal relationships and reduce omissions in date-event summarization. Extensive experiments, including quantitative analysis on three TLS benchmarks and human evaluation, demonstrate that NTS-CoT outperforms state-of-the-art baselines, effectively mitigating hallucinations and improving LLM-based TLS performance. Our source code is available at https://anonymous.4open.science/r/NTS-CoT .

Large language model (LLM) agents are increasingly expected to operate over long-term interactions, where information from past dialogues must be preserved and recalled to support future tasks. However, as interactions accumulate, the memory store grows without bound and fills with redundant entries that inflate storage cost and degrade retrieval by crowding out the most useful evidence. Furthermore, this is especially limiting on resource-constrained platforms with hard memory budgets, motivating us to formulate storage-budgeted memory management, the task of keeping an already constructed memory store within a fixed budget while preserving information useful for future interactions. To this end, we then propose MemRefine, an LLM-guided framework that, since surface similarity poorly reflects factual value, uses similarity only to propose candidate pairs and defers delete, merge, and preserve decisions to an LLM judge based on factual content, iterating until the budget is met. Across multiple memory frameworks and long-term conversation benchmarks, MemRefine consistently meets target budgets while preserving downstream performance and outperforming rule-based baselines under tight budgets.

Contrastively trained vision-language models like CLIP, have made remarkable progress in learning joint image-text representations, but still face challenges in compositional understanding. They often exhibit a "bag-of-words" behavior--struggling to capture the object relations, attribute-object bindings, and word order dependencies. This limitation arises not only from the reliance on global, single-vector representations for optimization, but also from the insufficient exploitation and modeling of the rich compositional information inherently present in paired image text data. In this work, we propose MACCO (MAsked Compositional Concept MOdeling), a framework that masks compositional concepts in one modality and reconstructs them conditioned on the full contextual information from the other, enabling the model to capture and align cross-modal compositional structures more effectively. To facilitate this process, we introduce two auxiliary objectives that jointly align and regularize masked features both inter-modally and intra-modally. Extensive experiments on five compositional benchmarks, along with in-depth analyses, demonstrate that our approach not only significantly enhances compositionality in VLMs but also improves their ability to capture syntactic structure and linguistic information. Additionally, the improved compositionality also benefits text-to-image generation and multimodal large language model. Code is available at https://github.com/hiker-lw/MACCO.

Holistic visual tokenizers are fundamental to unified multimodal models (UMMs) as they map diverse visual inputs into a unified representation space. In this paper, we present HYDRA-X, the first UMM that unifies image and video tokenization within a single Vision Transformer (ViT). Our design is driven by two core challenges: efficiently injecting spatiotemporal reconstruction capability into a native ViT, and embedding image- and video-level semantic awareness into the latent space. To address the first, comprehensive ablations reveal two key findings: (1) frame-level causal temporal attention suffices for visual reconstruction, whereas full spatiotemporal attention degrades it; and (2) hierarchical temporal compression substantially outperforms single-step alternatives. To tackle the second, we propose a lightweight decompressor that upsamples temporally compressed features under joint image-video teacher supervision, thereby enforcing complementary semantic structures within the compact latent space. Building on this holistic tokenizer, we further propose a principled improvement of the editing pipeline: source-target interaction should occur at the latent level inside the tokenizer rather than at the semantic level inside the LLM, substantially improving editing consistency and accelerating convergence. Instantiated at the 7B dense model, HYDRA-X achieves strong performance across image and video understanding and generation tasks, paving the way for future unified-tokenizer UMMs.

Computational creativity in Interactive Fiction faces a fundamental tension: Large Language Models (LLM) may produce creative narratives but struggle with world coherence, while symbolic systems ensure consistency but lack creative flexibility. We present IVIE (Incremental & Validated Interactive Experiences), a neuro-symbolic approach to generating complete and playable interactive fiction worlds from scratch. Building upon PAYADOR's neuro-symbolic framework, IVIE implements a four-stage incremental generation pipeline that delegates creative decisions--setting and character creation, puzzle design--to LLMs while grounding the world state through symbolic validation. The system generates worlds with interconnected locations, functional items, non-player characters, and coherent puzzles, all structured around a central goal-oriented architecture. Human evaluation shows the approach generates immersive, thematically coherent worlds with high player engagement. Results seem to indicate that the neuro-symbolic approach successfully balances flexibility with narrative coherence: symbolic validation grounds LLM generation without eliminating generative freedom. However, challenges remain: LLM inconsistencies occasionally bypass puzzle constraints, and objective validation gaps allow some structurally impossible goals. We identify key design considerations for future neurosymbolic interactive storytelling systems, particularly regarding LLM capabilities and their limitations.

Cloning camera motion from reference videos is an important task in video generation, as videos provide intuitive and precise control. Existing methods either directly use parametric representations that fail to handle multi-shot generation or synthesize cross-paired data, which suffer from data scarcity, resulting in poor performance in complicated camera motion cloning. To address these issues, we introduce a general camera motion representation that encodes cameras as grid motion videos. This camera grid represents the camera parameters visually and supports the integration of diverse trajectories for multi-shot video generation. Building upon this, we propose OmniDirector, a unified framework trained on a million-scale camera grid-video pairs that coordinates characters, actions, and cameras to provide director-level control for multimodal diffusion transformers. Furthermore, we design a novel hierarchical prompt expansion agent that harmoniously integrates different control signals by systematically describing camera motion and visual content through understanding signal relationships. Extensive experiments demonstrate the superior performance and outstanding controllability of our framework. Project page: https://ymlinfeng.github.io/OmniDirector.github.io/

Turn-taking in multi-party spoken conversations remains a fundamental challenge for voice-based agents, particularly under dynamic floor competition and varying user expectations. We propose ModeratorLM, a role-playing voice agent that conditions turn-taking behavior on an explicitly assigned role in multi-party settings. The system is built on a speech large language model operating in chunk-wise streaming manner. We further introduce a reasoning-augmented variant that incorporates chain-of-thought reasoning over conversational context and the assigned role. We construct RolePlayConv, a large-scale synthetic dataset of spoken multi-party conversations with diverse assistant roles. Experiments on real-world meeting data and RolePlayConv show improved turn-taking precision by over 40% and recall by more than 70%, while substantially reducing false-positive interruptions compared to non-role-conditioned baselines.

Multimodal Large Language Models (MLLMs) have shown promising reasoning capabilities in general domains, yet their performance remains limited in specialized settings such as healthcare, especially in multilingual and low-resource scenarios. This gap is critical in regions like rural India, where patients often express complex medical queries in native Indic languages and rely on multimodal inputs such as medical images. Existing English-centric MLLMs struggle to support such use cases, limiting equitable access to AI-driven healthcare assistance. To address this challenge, we introduce ArogyaBodha, a large-scale multilingual multimodal medical question-answer dataset constructed from eight heterogeneous sources, covering 31 body systems, six imaging modalities, and 21 clinical domains across English and seven major Indian languages. We further propose ArogyaSutra, an actor-critic-based multi-agent framework that integrates tool grounding with dual-memory mechanisms for step-wise, reasoning-aware decision making, and uses stored actor-critic simulation trajectories for distillation. Experiments show that our dataset and framework improve multilingual medical reasoning accuracy across all Indic languages, with ablations validating the contribution of each component. The source code and dataset are available at: https://iitp-cse.github.io/ ArogyaSutra/

Event-based vision has drawn increasing attention owing to its distinctive properties, including ultra-high temporal resolution and extreme dynamic range. Recent works have introduced it to video super-resolution (VSR) to enhance flow estimation and temporal alignment. In contrast, this paper shifts the focus of event signals from motion refinement to texture enhancement in VSR. We propose EvTexture++, the first event-driven framework dedicated to texture enhancement in VSR. It leverages high-frequency spatiotemporal details from events to improve texture recovery. EvTexture++ incorporates a customized texture enhancement branch, along with an iterative texture enhancement module that progressively exploits high-temporal-resolution event information for texture restoration. This enables gradual refinement of texture regions across iterations, yielding more accurate and detailed high-resolution outputs. Besides intra-frame texture recovery, large motions could degrade inter-frame temporal consistency, particularly in texture regions, leading to texture flickering. To mitigate this, we further exploit the continuous-time motion cues of events to enhance temporal consistency, introducing a temporal texture alignment module that estimates event-guided texture-aware flow for precise inter-frame texture alignment. Moreover, EvTexture++ is designed as a plug-and-play tool to flexibly boost the performance of existing VSR models. Experiments on five datasets demonstrate that EvTexture++ achieves state-of-the-art performance. When integrated into recent VSR models, it yields significant improvements, with gains of up to 1.55 dB in PSNR on the texture-rich Vid4 dataset. Code: https://github.com/DachunKai/EvTexture.

Retrieval-augmented generation (RAG) has become a standard mechanism for grounding language models in external knowledge, yet conventional retrieval based on lexical or semantic similarity is poorly suited for complex reasoning tasks: a semantically similar problem may demand an entirely different solution strategy, while a superficially different problem may share the same underlying reasoning pattern. We propose Retrieval-Augmented Reinforcement Fine-Tuning (RA-RFT), a post-training framework that teaches language models to reason by analogy. RA-RFT uses gold-relevance distillation to train a retriever that ranks contexts by expected reasoning benefit rather than semantic overlap, and then fine-tunes the policy model via reinforcement fine-tuning methods with retrieved analogous demonstrations, so the model learns to leverage reasoning traces under verifiable outcome rewards. We further analyze the diversity of retrieved contexts and find that reasoning-aware retrieval surfaces complementary solution strategies that provide distinct reasoning scaffolds for individual problems. Across challenging mathematical reasoning benchmarks, RA-RFT consistently outperforms standard reinforcement fine-tuning methods. For example, it improves AIME 2025 average@32 accuracy by 7.1 and 2.8 points over GRPO for Qwen3-1.7B and Qwen3-4B respectively -- suggesting that reasoning-aware retrieval is a complementary axis of improvement and orthogonal to advances in reward design or training curricula.

Recent progress in Large Multi-modal Models (LMMs) has enabled effective vision-language reasoning, yet the ability to video understanding remains constrained by suboptimal frame selection strategies, albeit with the rapid development of video-specialized LMMs. Prior works attempted to solve this with static heuristics or external retrieval modules to feed frame-level information, but these approaches often fail to capture visual cues grounded to the given user queries conflating raw visual dynamics with true semantic relevance. In this paper, we introduce ReFoCUS (Reinforcement-guided Frame Optimization for Contextual UnderStanding), the first framework to integrate online policy-gradient reinforcement learning into frame-level optimization for video-LLMs. ReFoCUS aims to learn a frame selection policy, leveraging reward signals derived from reference models to capture their underlying scoring behavior over frame combinations that best support temporally grounded responses. To efficiently explore the large combinatorial frame space, we employ an autoregressive and query-conditional selection architecture that ensures contextual consistency while reducing complexity. Our policy learning removes the need for explicit frame-level supervision, as it implicitly discovers optimal and semantically consistent frame compositions. ReFoCUS consistently improves reasoning accuracy across multiple video QA benchmarks, demonstrating the advantage of aligning frame selection with model-internal utility.

As large language models (LLMs) increasingly power conversational agents, understanding how they model users' emotional states is critical for ethical deployment. Inspired by emotion wheels, i.e., a psychological framework that argues emotions organize hierarchically, we analyze probabilistic dependencies between emotional states in model outputs. We find that LLMs naturally form hierarchical emotion trees that align with human psychological models, and larger models develop more complex hierarchies. We also uncover systematic biases in emotion recognition across socioeconomic personas, with compounding misclassifications for intersectional, underrepresented groups. Human studies reveal striking parallels, suggesting that LLMs internalize aspects of social perception. Beyond highlighting emergent emotional reasoning in LLMs, our results hint at the potential of using cognitively-grounded theories for developing better model evaluations.

As Large Language Models (LLMs) have reached human-like fluency and coherence, distinguishing machine-generated text (MGT) from human-written content becomes increasingly difficult. While early efforts in MGT detection have focused on binary classification, the growing landscape and diversity of LLMs require a more fine-grained yet challenging authorship attribution (AA), i.e., being able to identify the precise generator (LLM or human) behind a text. However, AA remains nowadays confined to a monolingual setting, with English being the most investigated one, overlooking the multilingual nature and usage of modern LLMs. In this work, we introduce the problem of Multilingual Authorship Attribution, which involves attributing texts to human or multiple LLM generators across diverse languages. Focusing on 18 languages -- covering multiple families and writing scripts -- and 8 generators (7 LLMs and the human-authored class), we investigate the multilingual suitability of monolingual AA methods in terms of their cross-lingual transferability, and the impact of generators on attribution performance. Our results reveal that while certain monolingual AA methods can be adapted to multilingual settings, significant limitations and challenges remain, particularly in transferring across diverse language families, underscoring the complexity of multilingual AA and the need for more robust approaches to better match real-world scenarios.

Large Language models (LLMs) have shown strong capabilities in code review automation, such as review comment generation, yet they suffer from hallucinations -- where the generated review comments are ungrounded in the actual code -- poses a significant challenge to the adoption of LLMs in code review workflows. To address this, we explore effective and scalable methods for a hallucination detection in LLM-generated code review comments without the reference. In this work, we design HalluJudge that aims to assess the grounding of generated review comments based on the context alignment. HalluJudge includes four key strategies ranging from direct assessment to structured multi-branch reasoning (e.g., Tree-of-Thoughts). We conduct a comprehensive evaluation of these assessment strategies across Atlassian's enterprise-scale software projects to examine the effectiveness and cost-efficiency of HalluJudge. Furthermore, we analyze the alignment between HalluJudge's judgment and developer preference of the actual LLM-generated code review comments in the real-world production. Our results show that the hallucination assessment in HalluJudge is cost-effective with an F1 score of 0.85 and an average cost of $0.009. On average, 67% of the HalluJudge assessments are aligned with the developer preference of the actual LLM-generated review comments in the online production. Our results suggest that HalluJudge can serve as a practical safeguard to reduce developers' exposure to hallucinated comments, fostering trust in AI-assisted code reviews.

Retrieval-Augmented Generation (RAG) extends large language models (LLMs) beyond parametric knowledge, yet it is unclear when iterative retrieval-reasoning loops meaningfully outperform static RAG, particularly in scientific domains with multi-hop reasoning, sparse domain knowledge, and heterogeneous evidence. We provide the first controlled, mechanism-level diagnostic study of whether synchronized iterative retrieval and reasoning can surpass an idealized static upper bound (Gold Context) RAG. We benchmark eleven state-of-the-art LLMs under three regimes: (i) No Context, measuring reliance on parametric memory; (ii) Gold Context, where all oracle evidence is supplied at once; and (iii) Iterative RAG, a training-free controller that alternates retrieval, hypothesis refinement, and evidence-aware stopping. Using the chemistry-focused ChemKGMultiHopQA dataset, we isolate questions requiring genuine retrieval and analyze behavior with diagnostics spanning retrieval coverage gaps, anchor-carry drop, query quality, composition fidelity, and control calibration. Across models, Iterative RAG consistently outperforms Gold Context, with gains up to 25.6 percentage points, especially for non-reasoning fine-tuned models. Staged retrieval reduces late-hop failures, mitigates context overload, and enables dynamic correction of early hypothesis drift, but remaining failure modes include incomplete hop coverage, distractor latch trajectories, early stopping miscalibration, and high composition failure rates even with perfect retrieval. Overall, staged retrieval is often more influential than the mere presence of ideal evidence; we provide practical guidance for deploying and diagnosing RAG systems in specialized scientific settings and a foundation for more reliable, controllable iterative retrieval-reasoning frameworks.

Transforming a large language model (LLM) into a vision-language model (VLM) can be achieved by mapping the visual tokens from a vision encoder into the embedding space of an LLM. Intriguingly, this mapping can be as simple as a shallow MLP transformation. To understand why LLMs can so readily process visual tokens, we need interpretability methods that reveal what is encoded in the visual token representations at every layer of LLM processing. In this work, we introduce LatentLens, a novel approach for mapping latent representations to descriptions in natural language. LatentLens encodes a large text corpus and stores contextualized token representations for each token in that corpus. Visual token representations are then compared to these contextualized representations and the top-nearest neighbor representations serve as descriptions of the visual token. We evaluate this method on 15 different VLMs, showing that commonly used methods, such as LogitLens, substantially underestimate the interpretability of visual tokens. With LatentLens instead, the majority of visual tokens are interpretable across all studied models and all layers. Qualitatively, we show that the descriptions produced by LatentLens are semantically meaningful and provide more fine-grained interpretations for humans compared to individual tokens. More broadly, our findings contribute new evidence on the alignment between vision and language representations and open up new directions for analyzing the latent representations of LLMs.

Omni-modal large language models (OLLMs) aim to unify multimodal understanding and generation, yet extending them to jointly produce speech and 3D facial animation remains largely unexplored despite its importance for natural human-computer interaction. A key challenge is the mismatch between the discrete semantic reasoning of LLMs and the dense temporal dynamics required for 3D facial motion. We propose Expressive Omni (Ex-Omni), an open-source model that augments OLLMs with native speech-accompanied 3D facial animation. Ex-Omni decouples semantic reasoning from temporal generation through a blendshape-aware speech unit generator and a blendshape decoder, where speech units provide temporal scaffolding and hidden speech representations carry facially relevant cues. We further introduce a unified token-as-query gated fusion (TQGF) mechanism for controlled semantic injection, as well as InstructS2SF-1200K, a dataset consisting of 1200K samples for pre-training. Extensive experiments show that Ex-Omni maintains competitive speech understanding and generation ability while achieving better audio-visual synchronization and lower face-generation latency than cascaded pipelines.

The rapid evolution of Large Language Models has catalyzed a surge in scientific idea production, yet this leap has not been accompanied by a matching advance in idea evaluation. The fundamental nature of scientific evaluation needs knowledgeable grounding, collective deliberation, and multi-criteria decision-making. However, existing idea evaluation methods often suffer from narrow knowledge horizons, flattened evaluation dimensions, and the inherent bias in LLM-as-a-Judge. To address these, we regard idea evaluation as a knowledge-grounded, multi-perspective reasoning problem and introduce InnoEval, a deep innovation evaluation framework designed to emulate human-level idea assessment. We apply a heterogeneous deep knowledge search engine that retrieves and grounds dynamic evidence from diverse online sources. We further achieve review consensus with an innovation review board containing reviewers with distinct academic backgrounds, enabling a multi-dimensional decoupled evaluation across multiple metrics. We construct comprehensive datasets derived from authoritative peer-reviewed submissions to benchmark InnoEval. Experiments demonstrate that InnoEval can consistently outperform baselines in point-wise, pair-wise, and group-wise evaluation tasks, exhibiting judgment patterns and consensus highly aligned with human experts.

Reinforcement learning has recently improved the reasoning ability of Large Language Models and Multimodal LLMs, yet prevailing reward designs emphasise final-answer correctness and consequently tolerate process hallucinations--cases where models reach the right answer while misperceiving visual evidence. We address this process-level misalignment with PaLMR, a framework that aligns not only outcomes but also the reasoning process itself. PaLMR comprises two complementary components: a perception-aligned data layer that constructs process-aware reasoning data with structured pseudo-ground-truths and verifiable visual facts, and a process-aligned optimisation layer that constructs a hierarchical reward fusion scheme with a process-aware scoring function to encourage visually faithful chains-of-thought and improve training stability. Experiments on Qwen2.5-VL-7B show that our approach substantially reduces reasoning hallucinations and improves visual reasoning fidelity, achieving state-of-the-art results on HallusionBench while maintaining strong performance on MMMU, MathVista, and MathVerse. These findings indicate that PaLMR offers a principled and practical route to process-aligned multimodal reasoning, advancing the reliability and interpretability of MLLMs.

Large Language Models (LLMs) reveal inherent and distinctive personas through dialogue. However, most existing persona discovery approaches rely on surface-level lexical or stylistic cues, treating dialogue as a flat sequence of tokens and failing to capture the deeper discourse-level structures that sustain persona consistency. To address this limitation, we propose a novel analytical framework that interprets LLM dialogue through bridging inference -- implicit conceptual relations that connect utterances via shared world knowledge and discourse coherence. By modeling these relations as structured knowledge graphs, our approach captures latent semantic links that govern how LLMs organize meaning across turns, enabling persona discovery at the level of discourse coherence rather than surface realizations. Experimental results across multiple reasoning backbones and target LLMs, ranging from small-scale models to 80B-parameter systems, demonstrate that bridging-inference graphs yield significantly stronger semantic coherence and more stable persona identification than frequency or style-based baselines. These results show that persona traits are consistently encoded in the structural organization of discourse rather than isolated lexical patterns. This work presents a systematic framework for probing, extracting, and visualizing latent LLM personas through the lens of Cognitive Discourse Theory, bridging computational linguistics, cognitive semantics, and persona reasoning in large language models. Codes are available at https://github.com/JiSoo-Yang/Persona_Bridging.git

Modern Vision-Language Models (VLMs) achieve strong semantic recognition, yet remain brittle on elementary spatial relations such as left of, on, behind, and between. One cause of this failure arises before language reasoning begins: the visual pathway may compress or discard critical 3D structural cues during feature extraction, so the language model receives image representations that are already insufficient for reliable spatial judgment. We introduce GeoWorld-VLM, a VLM-side distillation framework that transfers geometric structure from frozen camera-conditioned video world models into VLMs. GeoWorld-VLM fine-tunes only the image encoder and multimodal projector, aligning post-projector image features with intermediate world-model representations while leaving the main backbone frozen. Given images, a prompt, and a sampled camera trajectory, the world-model teacher converts static visual input into a synthetic multi-view spatial signal. Training combines spatial answer supervision, teacher-student feature alignment, and a preservation anchor to the original VLM. Since the language model remains frozen, GeoWorld-VLM preserves the original model's linguistic capabilities while attributing spatial improvements to the enhanced visual pathway. To evaluate the effectiveness and generality of the proposed method, we apply GeoWorld-VLM to two distinct VLM architectures and observe consistent improvements across both backbones. GeoWorld-VLM improves performance by approximately 4 percent on both the What'sUp and VSR benchmarks, suggesting that world-model-guided visual alignment generalizes across model structures and spatial reasoning datasets.

Detecting Schwartz values in political text is difficult because implicit cues often depend on surrounding arguments and fine-grained distinctions between neighboring values. We study when context and explicit moral knowledge help sentence-level value detection. Using the ValuesML/Touché ValueEval format, we compare sentence, window, and full-document inputs; no-RAG and retrieval-augmented settings with a curated moral knowledge base; supervised DeBERTa-v3-base/large encoders; and zero-shot LLMs from 12B to 123B parameters. The results show that more context is not uniformly better: full-document context improves supervised DeBERTa encoders by 3.8-4.8 macro-F1 points over sentence-only input, but does not consistently help zero-shot LLMs. Retrieved moral knowledge is more consistently useful in matched comparisons, improving each tested model family and context condition under early fusion. However, scaling from DeBERTa-v3-base to large and from 12B to larger LLMs does not guarantee gains, and simple early fusion outperforms the tested late-fusion and cross-attention RAG variants for encoders. Per-value analyses show that context and retrieval help most for socially situated or conceptually confusable values. These findings suggest that value-sensitive NLP should evaluate context, knowledge, and model family jointly rather than treating longer inputs or larger models as universal improvements.

Pre-trained language models (PLMs) have achieved strong performance in keyphrase extraction (KPE), largely due to their ability to generate rich contextualized representations. However, long-document KPE remains challenging because salient keyphrase evidence may be scattered across distant document sections that cannot be jointly captured within the limited context window of most PLMs. Although long-context large language models (LLMs) can process broader textual contexts, their computational cost limits their practicality for efficient and high-throughput KPE. To overcome this limitation, we propose an attention expansion mechanism that augments PLM token representations with information from surrounding out-of-context chunks using pre-trained word embeddings. The proposed mechanism expands the effective contextual scope of PLM-based KPE models without requiring full-document attention or expensive LLM-based inference. We evaluate our approach across five PLM backbones, including general-purpose, scientific, task-specific, and long-context encoders, using two training regimes and five benchmark corpora from scientific and news domains. Experimental results demonstrate that attention expansion consistently enhances KPE performance across all evaluation settings, outperforming state-of-the-art models and yielding notable improvements in F1 score. The improvements extend to domain-specific, task-specialized, and native long-context models, showing that the proposed mechanism provides complementary information rather than merely compensating for limited input length. These results establish attention expansion as an efficient and effective strategy for long-document KPE.

Video Large Multimodal Models have achieved remarkable progress in video understanding, yet they remain prone to hallucinations, where generated responses are not faithfully supported by the input video. In this paper, we propose MultiToP, a multimodal-context-aware visual token patching framework that mitigates hallucinations by refining unreliable visual tokens before language generation. MultiToP introduces a lightweight Visual Token Patcher to predict token-level replacement distributions and selectively substitute unreliable visual tokens with a dynamic global patch token. To train the patcher effectively, we further propose information-guided rank calibration, which uses answer-conditioned frame-level information cues derived from the backbone to guide token replacement. Combined with ground-truth answer supervision and sparsity regularization, MultiToP enables localized visual evidence refinement without modifying the original model. Extensive experiments demonstrate that MultiToP effectively reduces hallucinations on Vript-HAL with negligible inference overhead, improving the F1 scores of Qwen3-VL-4B-Instruct by 50.60% over the vanilla model. Meanwhile, MultiToP preserves general video understanding ability, yielding an 18.58% relative accuracy gain on ActivityNet-QA for Video-LLaVA-7B.

Closed-loop driving simulators typically populate their environments with non-ego traffic agents that behave largely the same way, produced either by rule-based traffic managers or by learned models trained toward a single behavioral mode. Recent work introduces style variation through post-hoc labels on observational data or LLM-inferred reward weights, but these signals act as proxies for what a style should reward rather than demonstrations of humans explicitly asked to drive in that style. We introduce PersonaDrive, a pipeline that conditions a vision-language-action (VLA) driving agent on retrieved demonstrations from a style-instructed human driving dataset, in which participants drive CARLA leaderboard routes under aggressive, neutral, and conservative instructions on a driver-in-the-loop rig. The pipeline has three stages: (i) offline triplet mining over per-style human driving data using a combined image-text similarity score; (ii) training a lightweight retrieval head that fuses frozen visual features with a small control encoder over per-style databases; and (iii) fine-tuning a single VLA backbone to treat retrieved context points as in-context behavioral demonstrations during waypoint prediction. At inference, the same backbone is conditioned on any style by swapping which per-style database the retrieval head queries, so selecting a style requires no per-style retraining while enabling human-style, style-diverse non-ego agents for closed-loop simulation. On Bench2Drive, PersonaDrive (no style) improves the driving score by 4.6% over SimLingo and 2.5% over HiP-AD, and under style conditioning attains the highest driving score in every style within a roughly 2% band (its weakest style surpassing the strongest baseline, DMW, by 5.4%), while average speed and acceleration rise by 18% and 25% from the conservative to the aggressive instruction.

Open-world mapless navigation from sparse language instructions requires resolving underspecified goals and inferring which environmental cues are relevant for reaching the goal. For instance, reaching an out-of-view destination may require interpreting ramps, signs, or detours that reveal where to go or which route to take. Prior works are limited by their reliance on known navigation factors and closed-set factor categories, or identify cues before motion planning and miss plan-dependent cues. We argue that pretrained Vision-Language Models (VLMs) can discover novel instruction-relevant cues, but require adaptation to focus on which cues matter and how they should influence motion planning. We realize these ideas in Foresight, a test-time framework in which a finetuned VLM alternates between proposing image-space motion plans and critiquing them using the language goal and visual context. Subsequent plans are conditioned on prior critiques, enabling iterative motion refinement before execution. To align plan critiques and refinements with open-set behavior preferences, we learn a reward model from human feedback and use it to post-train the VLM with reinforcement learning in the plan-critique loop. In offline evaluations and 6 real-world environments, Foresight improves average task success by 37% and reduces interventions per mission by 52% relative to state-of-the-art test-time reasoning and foundation-model baselines, while running in real-time on a Jetson AGX Orin. We will release code, data, and training details to support future work on test-time reasoning for robot motion refinement. Additional videos at: https://amrl.cs.utexas.edu/foresight

Autonomous long-horizon sidewalk navigation is essential for micro-mobility applications such as robotic food delivery and assistive electronic wheelchairs. Unlike autonomous driving on the road, long-horizon sidewalk navigation requires precise maneuvering through unpredictable sidewalk terrains and pedestrians, with a lightweight perception stack as minimal as a single monocular RGB camera. While imitation learning (IL) from demonstrations offers a practical solution, the resulting autopilot policy often suffers from compounding errors, a lack of social compliance on sidewalks, and deficiencies in counterfactual reasoning to handle complex situations. To address these challenges, we introduce FlowPilot, a mapless navigation policy that achieves robust and efficient long-horizon navigation performance using only a monocular RGB camera. We first propose to use anchored flow matching as an action representation for policy pre-training on large-scale robot fleet data and to capture the diverse, complex, multimodal distribution of sidewalk navigation behaviors. To bridge the gap between imitation and alignment, we further design a human-in-the-loop preference learning scheme to tune the policy on a small amount of human intervention data. It strengthens the model's counterfactual reasoning and social compliance on sidewalks. We evaluate FlowPilot through extensive simulation and real-world experiments in diverse sidewalk environments. FlowPilot achieves 42% success rate and 66% route completion in simulation, while FlowPilot-HP further improves real-world robustness and social compliance, reducing IR by 40.0% and NIR by 52.1% relative to the base model.

In this paper, we propose the Enhanced World Action Model (EWAM), a closed-loop online adaptation architecture built upon a pretrained and fully frozen Cosmos3 backbone network. Evaluated entirely under a zero-shot task protocol, EWAM is centrally focused on reducing the amount of additional deployment data required to adapt to new task layouts. Notably, no extra task-specific demonstration sets were introduced in any of the evaluations, and no fine-tuning was performed on the backbone network. Its performance gains stem entirely from an inference-time co-reasoning mechanism composed of four inserted lightweight neural layers: the Neural Experience Memory Layer located in the intermediate layers of the Diffusion Transformer (DiT) provides task-relevant execution context; the Neural Anomaly Detection Layer after the state prediction head monitors the divergence between predicted and actual states in real time; the Neural Policy Routing Layer dynamically selects direct execution, conservative replanning, or rollback recovery based on the anomaly severity; and the Neural Action Correction Layer refines the generated action chunks using execution diagnostics. Unlike naive feature fusion, the memory, anomaly detection, and correction modules are deeply integrated into the Cosmos3 forward path in a differentiable manner, with only the final routing decision being a discrete supervised one.

Recent reinforcement learning approaches have shown great promise in improving humanoid motion tracking performance and achieving fall recovery under disturbances. However, most existing works treat motion tracking and fall recovery as different tasks and require multi-stage training with specialized recovery rewards and/or separate recovery policies. Moreover, existing reinforcement learning-based methods often terminate training episodes immediately after severe tracking failures, limiting recovery-oriented exploration in unstable or fallen states. To address the above issues, we propose Stubborn, a streamlined and unified reinforcement learning framework to achieve robust humanoid motion tracking and fall recovery. Specifically, Stubborn uses an asymmetric Actor-Critic architecture and consists of three major components. First, a yaw-aligned tracking representation is adopted to reduce sensitivity to global drift and heading disturbances while preserving gravity-related balance information. Second, we introduce a Bernoulli-based probabilistic termination mechanism that enables the policy to encourage exploration of fall-recovery behaviors under varying failure modes. Third, we propose a probabilistic termination and tracking-error-driven strategy that dynamically reshapes the sampling distribution based on tracking performance, increasing the training efficiency for difficult motion segments and unstable states. Extensive comparisons with SOTA methods and ablation studies show that Stubborn achieved competitive performance, and the proposed probabilistic termination mechanism and adaptive sampling strategy contributed to the performance and robustness gains. For real-world demonstrations, please refer to https://aislab-sustech.github.io/Stubborn/.

Code-writing large language models (CodeLLMs) generate executable code policies for embodied agents by translating natural language goals and environmental constraints into structured control programs. However, policy generation in open-domain embodied environments suffers from two fundamental limitations: (i) delayed decoding caused by repetitive prefill computation over long prompts, and (ii) limited robustness due to fully generative decoding, which often produces API mismatches, missing safety guards, and unstable control logic. To address these limitations, we present FCGraft, a Functional Cache Grafting framework. FCGraft maintains a library of function-level validated code skeletons and their associated prompt-level Transformer key-value (KV) caches, and synthesizes new policies by retrieving relevant functions and grafting their KV caches when a new task is provided. Given retrieved function caches, FCGraft performs cache grafting via stitching, which composes cached function segments into a composite policy, and patching, which locally adapts only the necessary code regions to satisfy task-specific parameters and constraints with minimal additional decoding. By eliminating redundant prefill computation, this approach reduces generation latency, while reusing validated control structures improves robustness over prompt-level caching methods RAGCache, achieving 18.31% higher task success rate and 2.3x faster policy synthesis.

Distinguishing self from others is a prerequisite for social intelligence, yet humanoid robots that increasingly share workspaces with humans still lack this ability. Here we show that a humanoid robot can learn self-other distinction from proprioceptive-visual correspondence, without any identity labels or kinematic models. Once established, this distinction bootstraps a predictive self-model that maps joint configurations to three-dimensional body occupancy, capturing how the robot's body changes with action. In multi-agent scenes involving humans or morphologically identical robots, the system reliably identifies itself, learns a 3D self-model, and supports downstream tasks including target reaching, collision-aware motion planning, and human-to-robot motion retargeting. Together, these results outline a route toward bodily self-representation in robots that act and coordinate alongside others in shared physical environments. Project page: https://euron-zc.github.io/humanoid-self-model/.

Humor plays a central role in human social relationships, and recent advances in computational humor create new opportunities for integrating humor into human-robot interaction (HRI). While large language models (LLMs) can generate diverse forms of humor, it remains unclear how humor style, joke content, and language preference shape perceptions of robot-delivered humor in group settings. In this exploratory study, we employed a mixed factorial design in which participants evaluated AI-generated jokes delivered by a robot in a university classroom. We examined the effects of humor type (Affiliative, Self-Enhancing, Aggressive, Self-Defeating) and joke content (person-related vs. political) on perceived funniness and appropriateness, as well as preferred language. Results show that humor type significantly influences funniness, with Aggressive and Affiliative humor rated higher, while joke content primarily affects appropriateness, with person-related jokes preferred over political ones. Language preference was shaped by both joke content and participants' self-reported fluency and humor practices.

Real-time execution, enabled by asynchronous inference that ensures both smooth action trajectories and fast reactivity, is critical for realistic deployments of large-scale Vision-Language-Action models. However, recent work on real-time execution primarily focuses on variants of diffusion policies, even though it is more critical for autoregressive policies given their slower rollout speed in synchronous inference. In contrast, we demonstrate that autoregressive policies can achieve real-time execution by adjusting the tokenization horizon and applying constrained decoding, thereby guaranteeing strict latency bounds that enable multi-trajectory decoding to maximize performance. Across simulated and real-world environments, we find that the autoregressive policy consistently outperforms its equivalent-level flow-matching policy counterpart while achieving significantly improved task completion speeds from synchronous inference. Coupled with the inherent advantages of autoregressive policies, such as faster convergence and better generalizability in instruction-following, these results confirm that autoregressive policies can remain a competitive policy type supporting real-time execution.

Robust localization in unstructured environments, such as agricultural fields, is a critical challenge for autonomous systems. LiDAR sensors provide detailed 3D information about the environment and are invariant to lighting conditions. For this reason, LiDAR-based place recognition methods have gained significant attention. In this paper, we propose MinkUNeXt-VINE++, a novel approach that combines early fusion of heterogeneous LiDAR data from two sensors (Livox Mid-360 and Velodyne VLP-16) and a learned re-ranking strategy in inference time. This fusion leverages the strengths of each sensor to provide a more comprehensive representation of the environment. Additionally, the re-ranking approach is particularly important in repetitive environments, such as vineyards, as finding true positives is a major challenge. We evaluated our approach using the TEMPO-VINE dataset, which provides heterogeneous LiDAR data in vineyard environments across different phenological stages. Our results demonstrate that MinkUNeXt-VINE++ significantly improves place recognition performance compared to single-sensor approaches and state-of-the-art methods. MinkUNeXt-VINE++ achieves a 20% improvement in the Recall@1 metric compared to single-sensor approaches, and +30% including re-ranking. The code of our method is publicly available for reproduction.

Indoor vision-based localization systems are affected by detection noise, occlusions, and limited camera coverage, leading to uncertainty at multiple stages of the pipeline. While multi-camera data fusion is widely used to mitigate these issues, it is typically treated as a black-box component and evaluated solely end-to-end, obscuring its mechanistic contributions. To address this gap, this work investigates whether explicitly characterizing single-camera localization errors can be leveraged to calibrate and optimize multi-camera data fusion. We introduce a measurement-calibrated fusion approach that integrates component-wise error quantification, specifically isolating homography calibration, human detection, and motion tracking. A component-wise evaluation is conducted to quantify error contributions from homography calibration, human detection, and motion tracking. Experimental results show that data fusion improves localization accuracy compared to single-camera baselines. While measurement-calibrated fusion provides only limited improvement in absolute accuracy over standard fusion, it substantially reduces trajectory variance and improves motion smoothness, which are critical for applications requiring stable and continuous motion estimates. These results highlight the value of explicit error characterization when designing data fusion strategies for vision-based indoor positioning systems.

Articulated tool manipulation remains a major challenge in dexterous robotics due to the need to coordinate internal degrees of freedom and contact-rich interactions. While prior work has largely focused on rigid objects, articulated tool use remains underexplored because of its physical complexity and the difficulty of learning functional grasping and manipulation policies. We present Mana (Manipulation Animator), a general sim-to-real framework that reinterprets dexterous manipulation as an animation problem. Inspired by computer animation, Mana employs a coarse-to-fine pipeline that transforms procedurally-generated grasp keyframes into manipulation trajectories through motion planning and reinforcement learning. The data generation process is largely automatic, requiring only a few mouse clicks to specify functional affordances (<1 minute per tool). Across four articulated tools spanning different scales and joint types, Mana achieves zero-shot sim-to-real transfer for both grasping and in-hand manipulation, demonstrating a scalable approach to dexterous articulated tool use.

The field of Embodied AI is witnessing a rapid evolution toward general-purpose robotic systems, fueled by high-fidelity simulation and large-scale data collection. However, this scaling capability remains severely bottlenecked by a reliance on labor-intensive manual oversight from intricate reward shaping to hyperparameter tuning across heterogeneous backends. Inspired by LLMs' success in software automation and science discovery, we introduce \textsc{EmboCoach-Bench}, a benchmark evaluating the capacity of LLM agents to autonomously engineer embodied policies. Spanning 32 expert-curated RL and IL tasks, our framework posits executable code as the universal interface. We move beyond static generation to assess a dynamic closed-loop workflow, where agents leverage environment feedback to iteratively draft, debug, and optimize solutions, spanning improvements from physics-informed reward design to policy architectures such as diffusion policies. Extensive evaluations yield three critical insights: (1) autonomous agents can qualitatively surpass human-engineered baselines by 26.5\% in average success rate; (2) agentic workflow with environment feedback effectively strengthens policy development and substantially narrows the performance gap between open-source and proprietary models; and (3) agents exhibit self-correction capabilities for pathological engineering cases, successfully resurrecting task performance from near-total failures through iterative simulation-in-the-loop debugging. Ultimately, this work establishes a foundation for self-evolving embodied intelligence, accelerating the paradigm shift from labor-intensive manual tuning to scalable, autonomous engineering in embodied AI field.

Vision-Language-Action (VLA) models have emerged as a promising paradigm for general-purpose robotic control, with test-time scaling (TTS) gaining attention to enhance robustness beyond training. However, existing TTS methods for VLAs require additional training, verifiers, and multiple forward passes, making them impractical for deployment. Moreover, they intervene only at action decoding while keeping visual representations fixed-insufficient under perceptual ambiguity, where reconsidering how to perceive is as important as deciding what to do. To address these limitations, we propose SCALE, a simple inference strategy that jointly modulates visual perception and action based on 'self-uncertainty', inspired by uncertainty-driven exploration in Active Inference theory-requiring no additional training, no verifier, and only a single forward pass. SCALE broadens exploration in both perception and action under high uncertainty, while focusing on exploitation when confident-enabling adaptive execution across varying conditions. Experiments on simulated and real-world benchmarks demonstrate that SCALE improves state-of-the-art VLAs and outperforms existing TTS methods while maintaining single-pass efficiency.

Dexterous hands are essential for fine-grained manipulation, but their hardware designs vary substantially across embodiments. Differences in kinematics, joint definitions, and degrees of freedom make it difficult to define a shared state representation compared with parallel grippers. As a result, dexterous-hand data remains fragmented and difficult to use for joint training. In this work, we propose the Unified Dexterous Hand Model (UDHM), which maps human and robot hand states into a shared 22-DoF semantic interface. Based on UDHM, we introduce UniDexTok, a retargeting-free state tokenizer that learns embodiment-conditioned discrete tokens from standardized real joint states. UniDexTok provides a unified representation for heterogeneous dexterous hands without relying on retargeting or simulation data. Compared with the recent baseline UniHM, UniDexTok reduces MPJAE from 15.63 degrees to 0.16 degrees and MPJPE from 18.51 mm to 0.18 mm, corresponding to error reductions of 98.98% and 99.03%, respectively. These results improve reconstruction from centimeter-scale to sub-millimeter accuracy. Experiments further show that data from other embodiments improves target-embodiment reconstruction accuracy, demonstrating the benefit of cross-embodiment tokenization. UniDexTok also shows strong zero-shot and few-shot reconstruction ability when new dexterous hands are introduced.

Elite humanoid soccer shooting requires whole-body stability, high-impulse whole-body interactions, and accuracy to targets. Motion tracking-driven reinforcement learning (RL) provides stability in whole-body movement coordination, but a fixed reference makes it hard to adapt to varied ball positions and strike timings; in contrast, task reward-driven RL struggles to explore and discover valid kicks from scratch. We therefore introduce RoboNaldo, a three-stage motion-guided curriculum RL framework for high-impulse humanoid interaction. A single human-kick reference is used as a scaffold and progressively shifts optimization towards shooting performance. The curriculum first learns a stable whole-body kicking prior, then adapts the kick to free-kick settings where the ball is stationary at random positions, and finally extends it to moving-ball shooting through a locomotion-command and kick-trigger interface. A high-level heuristic planner controls this interface during training, while alternative high-level controllers can drive the same low-level policy at inference. In simulation, RoboNaldo demonstrates free-kick shot error 48.6% lower and shoot velocity 2.96x than prior work baselines. In real world on a Unitree G1 with onboard perception, RoboNaldo attains 0.73 m and 0.86 m average target shooting error from 3 m away in free-kick and moving-ball cases, accordingly. And the post-contact ball velocity reaches 13.10 m/s, which is 59-71% of reported professional open-play shot speed. Project page: https://opendrivelab.com/RoboNaldo.

Blind grasping with a dexterous hand is a crucial manipulation capability. Nevertheless, learning such tactile-only policies for real robots remains challenging due to the tactile sim-to-real gap and the limited expressiveness of sparse tactile signals. To bridge this gap, we propose a framework for tactile-only blind grasping that is deployable on a physical multi-fingered robotic hand. Our approach combines three key components. First, we introduce a Real2Sim tactile calibration pipeline that constructs a contact-calibrated digital-twin simulator capable of reproducing real tactile signals. Second, we improve the expressiveness of sparse tactile observations using a layout-aware tactile encoder, which incorporates sensor-geometry priors through self-supervised pretraining. Third, to improve generalization to unseen objects, we train object-specific reinforcement-learning experts in the calibrated simulator and aggregate their successful grasp trajectories into a tactile-conditioned Diffusion Policy. We evaluate our method on a physical LEAP Hand equipped with distributed tactile sensing across 10 seen and 10 unseen objects. The deployed policy achieves a 27\% real-world grasp success rate across all 20 objects, without real-world grasping demonstrations or visual input. Simulation ablations show that layout-aware tactile pretraining improves grasping performance, while sensing-level evaluations confirm that Real2Sim calibration increases the consistency of tactile contact events between simulation and hardware. Together, these results suggest that contact-event calibration, geometry-aware tactile representation learning, and diffusion-based policy aggregation provide an effective path toward tactile-only blind grasping on real dexterous robotic hands. Project page:Dex-Blind-Grasp.github.io.

Safety-relevant studies of language models, including alignment and jailbreaking evaluations and AI control protocols, often rely on prefilling model outputs. If AI models can recognize and act on the fact their prior assistant messages have been inserted or edited, the effectiveness and validity of these methods could be compromised. We investigate whether frontier language models can distinguish between tampered and untampered assistant-side context, a capability we call prefill awareness. To do so, we construct a binary preference benchmark across three prefill mechanisms, filtering for cases where models show consistent stances. We find that frontier models show substantial prefill awareness: Claude Opus 4.5 detects prefills opposing its preferences in 9-35% of cases with a 0% false positive rate when prompted; additionally, models often revert towards baseline behavior without explicitly reporting that the prefill was foreign. Controlled ablations later also show that detection and resistance rely on different cues, where stylistic mismatch mainly affects whether models flag a prefill as foreign, while preference mismatch mainly affects whether they revert toward their baseline answer. We also examine more realistic agentic settings such as misalignment-continuation evaluations and SWE-bench trajectories, where frontier models sometimes disavow prefilled assistant turns in ways that depend strongly on dataset, task success, and hidden formatting artifacts. Our results indicate that prefill awareness is already a substantial confound for some prefill-based methods. We recommend that model developers track this capability in frontier systems.

Agentic large language model systems that autonomously invoke tools, maintain persistent memory, and execute multi-step plans are increasingly deployed in public-facing domains, including government services, healthcare triage, and financial advising. We ask whether the frameworks used to build these systems provide architectural-level structural safety guarantees. Applying six containment principles derived from a compositional model of agentic architectures, we audit three dominant frameworks (LangChain, AutoGPT, and OpenAI Agents SDK) and find no native compliance in any of them. Memory integrity, a defense against one of the most prevalent vulnerability classes, is not observed in any of the three evaluated frameworks. We validate these findings empirically: in a simulated government benefits agent built on LangChain, a single memory-poisoning write induces persistent targeted corruption across all tested seeds and backends, increasing the wrongful denial rate for targeted applicants to 88.9%. Under a complex five-factor policy, the same attack preserves aggregate accuracy while increasing targeted wrongful denials by 3.5x, rendering the corruption difficult to detect through standard monitoring. We then introduce two lightweight containment mechanisms: a memory integrity validator and a policy gate, which eliminate both attack vectors with sub-millisecond overhead (<0.2ms per call). We conclude that the current agentic framework ecosystem may not yet meet secure-by-default expectations for public-facing deployments and outline priority architectural interventions to enable trustworthy deployment in high-stakes, socially impactful applications.

Large language models (LLMs) are increasingly used for tasks once reserved for trained researchers, including hypothesis generation, specification choice, and drafting conclusions. We argue that the reliability of AI-assisted research depends not only on model capability, but also on how cognitive labour is structured between humans and machines. We study this problem through Human-in-the-Loop Economic Research (HLER), a decision architecture based on pre-commitment, decision sequencing, accountability, and attention allocation. In a pre-specified 2*4 factorial experiment with 280 complete research runs across four datasets, an unconstrained multi-agent baseline produced critical failures in 72% of runs. Using the same underlying model, the same agent decomposition, and identical prompts for the shared reasoning agents, HLER reduced the failure rate to 16% by imposing three architectural commitments: LLMs reason but do not execute data work, data and estimation are handled deterministically, and three human decision gates bind the workflow. Fisher's exact test rejects equality of failure rates at p<0.001. Reliability gains were largest on the least publicly represented dataset, a Qing-dynasty population register, consistent with a task-based production model with Frechet-distributed output quality. An 80-run ablation suggests that deterministic computation and human gates contribute independently, with exploratory evidence of complementarity. We interpret HLER as a research harness rather than an autonomous AI scientist: it sharply reduces failures, makes residual weaknesses more visible, and prevents unreliable claims from being advanced as publication-ready outputs.

Large language models (LLMs) often hallucinate by generating factually incorrect or unfaithful content, posing significant risks to their safe use. Detecting such hallucinations is particularly challenging under the zero-source constraint, where no model internals or external references are available, and detection must rely solely on the textual query-answer pair. In this paper, we propose Human-like Criteria Probing for Hallucination Detection (HCPD), a paradigm that emulates the multi-faceted reasoning of human evaluators. Its core is a Human-like Criteria Probing (HCP) mechanism, in which a LLM agent adaptively decomposes its judgment into a weighted set of interpretable criteria and aggregates criterion-specific scores into a final truthfulness measure. To achieve this adaptive capability, we introduce a reward-based alignment scheme using only weak supervision from semantic consistency. At inference, we employ a multi-sampling aggregation strategy to ensure robust decisions while preserving full interpretability. We further provide theoretical analysis supporting the reliability of our approach. Extensive experiments show that HCPD consistently outperforms state-of-the-art baselines, offering an effective and explainable solution for zero-source hallucination detection. Code is available at https://github.com/TRISKEL10N/HCPD.

As AI systems are deployed in high-stakes ethical contexts such as healthcare triage, autonomous vehicle control, and employment screening, formal methods for evaluating their robustness against adversarial manipulation of ethical reasoning remain underdeveloped. This paper introduces the Ethical Robustness Testing System (ERTS), a closed-pipeline framework that: (1) encodes ethical dilemmas into a 22-dimensional Ethical Consequence Space (ECS) grounded in established ethical theory; (2) applies 17 semantic perturbation functions subject to 6 validity constraint classes including a novel semantic coherence constraint; (3) measures decision deviation via a 4-component Ethical Instability Index (EII); and (4) produces domain-adaptive pre-deployment robustness assessment verdicts. We evaluate 4 structured baseline models and 2 production LLMs (Gemini 2.0 Flash and Llama 3.2) across 50 ethical scenarios spanning 8 deployment domains, generating 1,500 adversarial test cases. Results demonstrate that only 33% of models achieve assessment clearance, with the local Llama-3.2 model proving particularly vulnerable to fairness corruption and information degradation attacks (ERS = 0.737). To the best of our knowledge, no existing framework combines a bounded ethical consequence space, semantic coherence constraints, and domain-adaptive assessment in a single adversarial testing pipeline.

Shielded reinforcement learning is typically presented as a runtime safety mechanism that compiles temporal-logic specifications into automata restricting an agent's actions. We argue this is the wrong product. The same automata-theoretic machinery -- specification compilation, product game construction, attractor computation, and winning-region extraction -- is better read as a design-time analytical instrument whose outputs are structural insights about a system rather than runtime constraints on a deployed agent. We instantiate this through a constrained two-player safety game for network defense. The two specifications are enforced asymmetrically: the defender specification defines the unsafe region of the game, whereas the attacker specification restricts the adversary's legal actions during attractor computation. Solving the game yields a defensibility verdict -- a formal certificate that a topology-specification pair is or is not defensible -- with the associated winning region and shield. Beyond the binary verdict, we derive topology-level metrics from the attractor structure and combine them with post-convergence behavior from shield-constrained adversarial multi-agent reinforcement learning. Together these form a defensibility fingerprint capturing both a network's formal safety properties and its operational behavior under adaptive play. A what-if analysis shows that formal defensibility and operational effectiveness capture distinct aspects of security: small architectural changes can produce large shifts in operational outcomes while leaving formal safety margins nearly unchanged. Shield synthesis is thus most valuable not as a deployment mechanism for safe agents, but as a framework for answering architectural questions about whether, where, and how a system can be defended. The defensibility verdict is the output, not the safe policy.

The rapid global expansion of artificial intelligence regulation has generated, across multiple jurisdictions, a demand for legal expertise dedicated to AI that the market has addressed in a fragmented manner. Data protection officers extend their remit beyond data protection law; privacy lawyers reposition themselves toward AI; compliance officers add AI chapters to their existing manuals. This paper argues that none of these adaptive responses adequately covers the professional space opened by the emerging global AI regulatory landscape, of which the EU Artificial Intelligence Act (Regulation (EU) 2024/1689) is the most comprehensive instance, alongside the Council of Europe Framework Convention on AI, the United States executive and sectoral framework, and analogous initiatives in the United Kingdom, Canada, Brazil, China, Japan, Singapore, and beyond. A distinct professional profile is required: the AI Legal Specialist, conceived as a jurist -- understood broadly to encompass any professional with advanced legal training -- operating at the intersection of legal interpretation and AI governance. The profile is juridically autonomous: it derives its existence from the structure of regulatory obligations generated wherever AI is subject to substantive regulation, rather than from any technical standard or the extension of adjacent roles. The paper provides a juridically grounded definition of the profile, argues for its autonomy from adjacent figures and international standards, proposes a reference competence architecture aligned with the European e-Competence Framework (e-CF, EN 16234-1) as a methodological choice, and articulates the conditions for its operational measurement through key performance indicators. The contribution is intended as a foundation for international standardization of the profile and as a reference for practice, curricula, and adoption across jurisdictions.

The rapid integration of artificial intelligence (AI) into critical infrastructure including healthcare, finance, energy, and defense, offers transformative benefits but also conflicts with evolving regulatory and governance frameworks. This paper presents a systematic literature review (SLR) to examine the challenges of balancing AI compliance and technological innovation across critical infrastructure sectors. The review follows established SLR guidelines to extract and synthesize insights from peer-reviewed articles, report, and institutional sources published between 2020-2025. The study identifies three interrelated challenges: fragmented regulations, excessive compliance burdens for smaller to medium enterprises (SMEs), and misaligned governance models. To address these challenges, the study highlights practical governance strategies, including risk-tiered regulation, compliance by design, and explainable AI, to support scalable and trustworthy AI deployment in critical sectors. Key contributions include a concise mapping of core AI-governance challenges and a conceptual diagram illustrating their overlap, as well as actionable strategies for policymakers and practitioner to harmonize oversight with innovation.

Muse Spark is the latest large language model developed by Meta. In this report, we first present evaluations for catastrophic risk domains under Meta's Advanced AI Scaling Framework, along with the evidence that informed our launch decision. We then discuss additional considerations, such as Muse Spark's broader content safety and behavioral profile, that are relevant to overall safety but fall outside the catastrophic risk domains governed by the Framework. Our preparedness results covering Chemical and Biological, Cybersecurity, and Loss of Control risks assess Muse Spark's deployment within Meta AI as presenting acceptable levels of residual risks under our Advanced AI Scaling Framework. We conducted a broad set of evaluations targeting dual-use and high-risk capabilities across these catastrophic risk domains. Those evaluations identified elevated risks prior to mitigations, with Chemical and Biological capabilities assessed as likely reaching the "high risk" category under the Advanced AI Scaling Framework before safeguards were applied. We have implemented a multi-layered set of mitigations that address the identified risks, and Muse Spark demonstrates state-of-the-art refusal across a range of benchmarks related to hazardous workflows in chemistry and biology. We therefore release Muse Spark as the underlying model of Meta AI.

The increasing encroachment of artificial intelligence (AI) on social life raises significant risks for society, particularly within the infospheres created and controlled by companies such as Google, Facebook, Apple, and Amazon. This article examines these risks through an in-depth analysis of Facebook's content moderation regime, which is already partially governed by algorithms. We argue that the idea of ethical engineering, often proposed in the literature as a solution to the governance challenges posed by AI, is inadequate for several reasons. In response, we develop an alternative framework, which we term "algorithmic constitutionalism." Our approach rests on three pillars: (a) a layered architecture consisting of two levels of code: (i) an operative or object level and (ii) a meta level designed to protect the system's core principles from algorithmically initiated change; (b) algorithmic meta-reasoning, which enables the system to operate simultaneously at both levels so that it can monitor, verify, and potentially correct in real time operations at the object level that depart from principles protected at the meta-code level; and (c) correction through deliberation. The article elaborates the concept of algorithmic constitutionalism and demonstrates how it may be applied to Facebook's content moderation regime. As part of this analysis, we examine the tension between societal constitutionalism and algorithmic constitutionalism. Paradoxically, attempts to subject AI systems to external deliberative control may also enable AI agents to intervene in that process, potentially undermining its purpose. The article concludes by considering the implications of this argument for the European Digital Services Act, which entered into force in October 2022.

Large language model (LLM) answer engines are increasingly used for information seeking, shifting visibility from ranked lists to synthesized answers. This enables Generative Engine Optimization (GEO), which targets LLM answer engines' evidence pool and generation. We analyze the search engine optimization (SEO) to GEO transition to identify two risks: (i) concentrated influence from low contestability and system sensitivity, and (ii) undisclosed commercial influence embedded in evidence and reasoning. We then formalize a general GEO pipeline to locate where optimization acts and compare academic and industry practices, revealing a third risk: (iii) academic-industry blind spots driven by visibility and evaluation asymmetries between offline setups and deployed systems. This position argues the need for answer-level governance and measurement: stronger contestability, high-precision disclosure, black-box auditing of material influence, and deployment-aligned metrics for exposure persistence.

At present, loss of control risks have gained much prominence in public discussion, particularly in relation to AI, with extensive discourse present among academics, frontier labs, and even governments. However, in the existing literature, the concept seems to rest on surprisingly weak foundations, where even those that discuss loss of control extensively do not first establish what control is and what exactly is being lost. Our paper aims to address these gaps. We establish a working definition of control by anchoring it to the "setting and getting of goals". Then, we discuss various aspects of control, built on foundational concepts from related fields like cybernetics, management control, and control theory. This includes who (or what) can be in control, and the things they require to be in control, such as the ability to set goals, having a functional control loop, having requisite variety, and having sufficient goal alignment. Once a framework for control is established, we then discuss how control can be lost, how AIs can contribute to such loss of control, and offer relevant recommendations for how one can maintain control. One interesting consequence of our work is that humanity, as individuals and as groups, can lose varying degrees of control as a result of AI behaviour that is far below the level of superintelligence; the potential for loss of control scenarios (as we define them) already exist, and have existed for a long time.

Retrieval-augmented generation (RAG) agents increasingly run with persistent memory that accumulates across user sessions. This creates a new attack surface: an adversary interacting only through normal channels can inject crafted memories that, once retrieved, steer the agent's responses for future users, without touching model weights or code. We call this Multi-Session Memory Poisoning (MSMP) and show that no existing defence certifies against it; static-corpus defences (RobustRAG, ReliabilityRAG) assume a fixed knowledge base, and heuristic filters are bypassed by fluent enterprise-style text. We present Signed Memory with Smoothed Retrieval (SMSR), the first defence with a certified robustness bound for this setting. Component 1 adds HMAC-SHA256 provenance at write time, blocking unsigned injection. Component 2 applies randomised memory ablation with verdict-based majority voting at query time, bounding the influence of authenticated adversaries. We prove that no provenance-free retrieval-time filter can certify against adaptive injection, derive a hypergeometric certificate for Component 2, and formalise the Consistent Minority Effect, whereby a consistent adversarial answer wins string-based voting as a numerical minority while verdict-based voting removes it. Across 15 enterprise scenarios (3,150 repeated trials), Component 1 cuts attack success from 93-100% to 0% for all unsigned variants. For an authenticated adversary with a single injection, Component 2 holds success to 8.0% (95% CI [5.8, 10.9], n=450), below the certified worst case. In an end-to-end query-only attack where the agent itself writes the poison rather than it being pre-seeded, SMSR reduces success from 65.3% to 5.3% (n=150, non-overlapping CIs) on a live agent stack. Clean-query utility is 90% (Component 1) and 85% (combined).

Large Language Models (LLMs) are rapidly evolving into agentic systems that interact with external tools and environments, introducing new security risks such as indirect prompt injection attacks through untrusted external sources. Existing defenses mainly focus on blocking malicious content at inference time, and current red-teaming methods primarily optimize attack success. As a result, developers have limited visibility into how latent prompt injections emerge and propagate through agents. We propose PI-Hunter, an automated agentic auditing framework for proactive vulnerability exposure in LLM agents. PI-Hunter constructs realistic source-aware test cases and iteratively evolves them through feedback-driven exploration to induce agents to retrieve and reveal latent malicious instructions embedded within external environments. Extensive experiments across multiple benchmarks, agent architectures, attacks, and defenses demonstrate that PI-Hunter substantially improves vulnerability exposure and attack-surface coverage over strong automated red-teaming baselines, while remaining effective under existing prompt injection defenses.

While real-world applications of reinforcement learning (RL) are becoming increasingly popular, the security of RL systems deserve more attention and exploration. In particular, recent work has revealed that RL agents are vulnerable to backdoor attacks, where a victim agent behaves normally under standard conditions but executes malicious actions when a specific trigger is activated. Existing backdoor defenses for RL either require access to the agent's internal parameters, operate only at the model or trajectory level, or are limited to specific attack types. To ensure the security of RL agents, we propose \texttt{PolicyGuard}, a \textit{test-time step-level} backdoor defense which leverages Gaussian Process (GP) posterior variance and adapts pseudo trajectories to enable uncertainty computation for individual time step. Besides, we also provide theoretical foundations to explain the efficacy of GP posterior variance. Extensive experiments across seven RL games demonstrate that PolicyGuard achieves state-of-the-art detection performance in most cases, with average AUROC of 0.856 for perturbation-based attacks and 0.859 for adversary-agent attacks.

Model fingerprinting, embedding user-specific identifiers (fingerprints) into generated outputs, has recently emerged as a popular solution to protect the intellectual property rights (IPR) of generative text-to-image (T2I) models and prevent unauthorized redistribution. In this work, we reveal a previously unexplored systematic vulnerability in existing generative model fingerprinting methods: they lack robustness against collusion attacks, where multiple attackers combine their models to remove or obscure the fingerprints. To address this issue, we take the first step towards a robust fingerprinting method for T2I models with anti-collusion capabilities. The proposed method encodes strings of bits, namely fingerprints, into the coefficients of a personalized normalization module (PNM) incorporated into T2I models, so that fingerprints can be reliably recovered from any generated image. To defend against collusion attacks and prevent unauthorized model redistribution, we introduce an anti-collusion mechanism based on lossless function-invariant parameter transformations. This mechanism significantly degrades the image generation quality of colluded models, making them effectively unusable. Moreover, our method allows developers to efficiently create multiple copies of fingerprinted T2I models by reparameterizing the PNM without the need for retraining. We also introduce a worst-case optimization strategy to improve robustness against model-level attacks. Our experiments demonstrate that the proposed method achieves high fidelity and robustness across multiple T2I image generation and editing tasks, with fingerprint extraction accuracy exceeding 99.5%. Compared with existing methods, our method demonstrates, for the first time, a notable proactive robustness to collusion attacks by significantly increasing the FID of colluded models.

Interfacing Artificial Intelligence (AI) with democracy is one of the most profound challenges of our times. On the one hand, AI comes with opportunities to overcome long-standing challenges in democracy, such as low participation in deliberative and voting processes with poor representation of people. On the other hand, new risks arise from AI algorithms that are privacy-intrusive, biased, manipulative, spread misinformation and influence election results. Moving beyond the over-simplistic question of whether AI is good or bad for democracy, the Handbook on Democracy in the Era of Artificial Intelligence asks instead: how to upgrade democracies and the principles they are built on, using AI? How to engage with AI and on what terms? Which new values and design principles are required to build democratic resilience? In 34 chapters by 59 authors across the world from different disciplines, we explore how AI can empower collective intelligence for democracy (Part 1) and what is the future of deliberative democracy using large language models and social media (Part 2). We also illustrate the role of AI for building resilient self-governance systems (Part 3) and the challenges of transforming democracy in the age of AI (Part 4). We conclude with broader perspectives (Part 5) that re-imagine the interplay of democracy and AI.

The UK government has adopted a pro-AI stance to help transform public service delivery in the face of severe financial pressures, but the path to translate this vision into responsible AI practice remains ill-defined. While UK policy is often set at the national level, local authorities are responsible for most public service delivery, and the rapid advance of AI-first narratives in the public sector is exposing fault lines in knowledge and practice at this national-local interface. This paper examines how responsible AI is interpreted and implemented at the interface between the UK's central government and local authorities, taking the high-stakes area of Special Educational Needs and Disabilities (SEND) as a case study. We present a thematic analysis of 17 semi-structured interviews with policymakers, practitioners, and third-sector professionals to identify barriers and enabling conditions for responsible AI where national policy meets local practice. We identify five interconnected challenges facing local authorities: shadow usage of AI and data privacy risks, market-government asymmetry in AI provision, insufficient workforce readiness, a lack of standardised definitions and measurements, and gaps in human accountability. For each, participants proposed actionable steps, from strengthening data protection frameworks and rebalancing the market-government relationship to enhancing workforce capacity. Our examination of SEND brings these challenges into sharper focus, showing how high-stakes decisions affecting vulnerable children and families intensify tensions around accountability, fairness, and human oversight, exposing the limits of a principle-based regulatory approach. We argue that responsible public sector AI requires both national policy adjustments and structural reforms to institutional capacity, values, and governance mechanisms at the local level.

This paper examines how algorithmic accountability in Canada's visa system is articulated institutionally and experienced by applicants across borders. We analyzed Immigration, Refugees and Citizenship Canada (IRCC)'s Algorithmic Impact Assessment (AIA) for the temporary resident visa (TRV) triage system using the algorithmic decision-making adapted for the public sector (ADMAPS) framework and analyzed Reddit discussions among applicants using a mixed-methods approach. We show that while institutional artifacts emphasize transparency, procedural safeguards, and bounded impacts, applicants engage in collective sensemaking to interpret opaque decisions, often relying on peer knowledge amid uncertainty. We identify three asymmetries between how institutional accountability is structured and how people perceive the process: epistemic asymmetry in access to decision logic, jurisdictional asymmetry in exposure shaped by geopolitical positioning, and temporal--relational asymmetry in how waiting and uncertainty are experienced. We emphasize why it is important to shift attention from institutional design to the uneven distribution of experiences with public-sector algorithmic governance. Together, these contributions demonstrate how algorithmic governance systems in the context of transnational migration produce structured asymmetries not captured by institutional disclosure frameworks, and how extending ADMAPS can account for those uneven translations of accountability.

Nowadays, the autonomous execution of cyberattacks capable of causing substantial real-world harm is widely regarded as one of the critical red lines that frontier AI systems must not cross. Within this broader red-line scenario, autonomous penetration represents a core enabling capability and subtask: the ability of LLM-powered AI systems to independently conduct adversarial operations against a target server without human intervention, identify and exploit vulnerabilities, and obtain unauthorized access or control. A growing body of work has sought to assess the autonomous penetration capabilities of AI systems. However, existing evaluations often employ opaque methodologies, rely on unrealistic or overly simplified penetration-testing scenarios, or provide LLMs with excessive prior knowledge and task-specific guidance, and cannot accurately capture the extent to which modern AI systems can autonomously perform this core capability within broader high-impact cyberattack scenarios. To address these limitations, we construct a new autonomous penetration evaluation framework consisting of two components: target servers and agent scaffolding. Specifically, on the target-server side, we design two levels of target environments based on the number of secure services without known vulnerabilities deployed alongside a vulnerable service: Tier~1 (one secure service) and Tier~2 (three secure services), resulting in a total of 300 target servers. Meanwhile, the agent scaffolding adopts a general-purpose agent architecture equipped with a set of general-purpose cybersecurity tools, without any target-specific prior knowledge. We evaluate 19 open-weight and proprietary LLMs, and find that current models achieve penetration success rates ranging from 10.7% to 69.3%. Moreover, we observe that autonomous penetration capability continues to improve alongside advances in overall model capability.

Web agents driven by large language models (LLMs) are increasingly deployed in real-world environments, where they operate over untrusted web content and execute actions with direct consequences. This makes them vulnerable to prompt-injection attacks, in which seemingly benign content embeds adversarial instructions that manipulate agent behaviour. Existing security benchmarks adopt an \textit{attack-centric} perspective, focusing on the technical feasibility of injections while overlooking the nuanced distribution of resulting harms. In practice, however, prompt-injection risk is victim-dependent: a single exploit can produce asymmetric consequences for different stakeholders, and the same attack pattern may exhibit substantially different effectiveness depending on whom it targets. To capture these properties, we introduce \textbf{\sysname}, a \textit{stakeholder-centric} benchmark to systematically categorize and attribute harm in real-world web agent systems. It distinguishes between affected entities (e.g., user, seller, platform), decomposes the attacks into concrete objectives, and evaluates each case with complementary outcome- and process-level metrics. Our results reveal substantial and heterogeneous vulnerabilities: not a single attack objective is reliably resisted by current agents, and failures distribute across qualitatively distinct modes ranging from \emph{stealthy parasitism} (attack succeeds without disrupting the user's delegated task) to \emph{misaligned disruption} (task disrupted without attack success) and \emph{compounded failure} (both adversarial objective and task integrity simultaneously violated). These patterns are missed by conventional evaluation, highlighting the need for stakeholder-aware assessment of LLM-based agents in real-world deployments. Benchmark is available at https://github.com/StakeBench/SBC.

Language operates as a mechanism of both marginalization and resistance, especially for minority communities navigating insensitive and harmful speech online. As content moderation increasingly depends on large language models (LLMs), concerns arise about whether these systems can recognize culturally insensitive speech-language that disregards or marginalizes the cultural and religious perspectives of historically underrepresented communities, often through implicit erasure, misrepresentation, or normative framing, rather than overt hostility. Focusing on Bangladesh's Hindu and Chakma communities -- the country's largest religious and Indigenous ethnic minorities, respectively -- this paper investigates the epistemic limits of LLM-based moderation systems and explores methods for incorporating minority perspectives. We co-created a culturally grounded corpus of insensitive speech with community members and integrated their narratives into moderation pipelines using retrieval augmented generation (RAG). Our tool, Mod-Guide, improves LLM sensitivity to minority viewpoints by leveraging contextual cues derived from lived experience. Through mixed-method evaluations involving both minority and majority participants, we demonstrate that RAG-enhanced moderation responses are more contextually accurate and perceived differently across ethnic lines. This work advances research in human-computer interaction, AI ethics, and social computing by foregrounding restorative justice and hermeneutical inclusion in the design of content moderation systems.

Search-augmented LLMs increasingly mediate everyday consumer recommendations by retrieving live web content. This creates a new risk: generative recommenders may consume polluted web content, such as fake reviews and promotional pages crafted to mislead recommendations. We ask: to what extent do search-augmented LLMs become unwitting promoters of fake products when consuming polluted retrieval results? To answer this, we introduce FORGE (Fake Online Recommendations in Generative Environments), a benchmark for measuring fake-product promotion under controlled web-content pollution. Given an upstream search result, FORGE locally rewrites real products in retrieved web pages into fake ones to simulate web-content pollution, and measures how often the LLM recommends the fake product. FORGE covers 225 real-world products across 15 categories and 5 consumer scenarios. Across 12 commercial and open-weights LLMs, all models are vulnerable: a single polluted page yields fooled rates of up to 27%, while the full top-3 replacement raises this to 73.8%. Vulnerability varies substantially across categories, increasing when models lack stable prior knowledge of the relevant products. Reasoning does not mitigate this vulnerability; instead, it often generates spurious social proof to justify false recommendations. We evaluate three defenses: skepticism prompting and consensus filtering (over model priors or cross-document evidence). Skepticism can exacerbate vulnerability, much like reasoning, while filtering risks suppressing legitimate products. We release FORGE at https://github.com/leoluolol/forge-benchmark.

Large language models increasingly function as artificial reasoners: they evaluate arguments, assign credibility, and express confidence. Yet their belief-forming behavior is governed by implicit, uninspected epistemic policies. This paper argues for an epistemic constitution for AI: explicit, contestable meta-norms that regulate how systems form and express beliefs. Source attribution bias provides the motivating case: I show that frontier models enforce identity-stance coherence, penalizing arguments attributed to sources whose expected ideological position conflicts with the argument's content. When models detect systematic testing, these effects collapse, revealing that systems treat source-sensitivity as bias to suppress rather than as a capacity to execute well. I distinguish two constitutional approaches: the Platonic, which mandates formal correctness and default source-independence from a privileged standpoint, and the Liberal, which refuses such privilege, specifying procedural norms that protect conditions for collective inquiry while allowing principled source-attending grounded in epistemic vigilance. I argue for the Liberal approach, sketch a constitutional core of eight principles and four orientations, and propose that AI epistemic governance requires the same explicit, contestable structure we now expect for AI ethics.

Detecting when a language model is wrong without ground truth labels is a fundamental challenge for safe deployment. Existing approaches rely on a model's own uncertainty -- such as token entropy or confidence scores -- but these signals fail critically on the most dangerous failure mode: confident errors, where a model is wrong but certain. In this work we introduce cross-model disagreement as a correctness indicator -- a simple, training-free signal that can be dropped into existing production systems, routing pipelines, and deployment monitoring infrastructure without modification. Given a model's generated answer, cross-model disagreement computes how surprised or uncertain a second verifier model is when reading that answer via a single forward pass. No generation from the verifying model is required, and no correctness labels are needed. We instantiate this principle as Cross-Model Perplexity (CMP), which measures the verifying model's surprise at the generating model's answer tokens, and Cross-Model Entropy (CME), which measures the verifying model's uncertainty at those positions. Both CMP and CME outperform within-model uncertainty baselines across benchmarks spanning reasoning, retrieval, and mathematical problem solving (MMLU, TriviaQA, and GSM8K). On MMLU, CMP achieves a mean AUROC of 0.75 against a within-model entropy baseline of 0.59. These results establish cross-model disagreement as a practical, training-free approach to label-free correctness estimation, with direct applications in deployment monitoring, model routing, selective prediction, data filtering, and scalable oversight of production language model systems.

Distributed collaborative intelligence (DCI), encompassing edge-to-edge architectures, federated learning, transfer learning, and swarm systems, creates environments in which emergent risk is structurally unavoidable: locally correct decisions by individual agents compose into globally unacceptable behavioral trajectories under uncertainty. Existing approaches such as constrained optimization, safe reinforcement learning, and runtime assurance evaluate acceptability at the level of individual actions rather than across behavioral trajectories, and none addresses the multi-participant, uncertainty-laden nature of DCI deployments. This paper introduces mechanical conscience (MC), a novel concept and simplified mathematical framework that operationalizes trajectory-level normative regulation for both single-agent and distributed intelligent systems. Mechanical conscience is defined as a supervisory filter that minimally corrects a baseline policy's actions to reduce cumulative deviation from a normatively admissible region, while accounting for epistemic uncertainty. We introduce associated constructs, conscience score, mechanical guilt, and resonant dependability, that provide an interpretable vocabulary and computable governance signals for this emerging field. Core theoretical properties are established: admissibility equivalence, existence of optimal regulation, and monotonic deviation reduction. Illustrative results demonstrate that MC-regulated agents maintain trajectory-level normative acceptability where conventional controllers drift outside admissible bounds, and that the framework naturally extends to suppress interaction-induced emergent risk in multi-agent DCI settings.

As autonomous and agentic AI systems scale in robotic and human-machine environments, managing hallucination and persistent but unjustified action remains an open challenge. Rather than attributing these failures solely to model or alignment limitations, this paper explores the architectural vulnerability of unbounded autonomy - the presumption that an agent should continue operating regardless of rising uncertainty. It introduces a theory of managed autonomy that defines intelligent behavior through the formal capacity to detect epistemic drift, suspend reasoning, attempt recovery, and ultimately surrender control when reliability diminishes. We instantiate this theory via the SMARt (Self-Managing Multi-tier Autonomous Reasoning with Regulated/Revoked transitions) model, a four-layer framework featuring Stable, Meta-cognitive, Assisted, and Regulated states. By developing a timed, guarded Petri net formulation, we establish theoretically bounded properties for the system, demonstrating how architecture can formally mandate escalation, constrain invalid outputs, and ensure governance reachability under specified conditions. We further analyze how incorporating domain-specific trigger sets across varied operational settings (e.g., healthcare, robotics, etc.) can systematically preserve safety, assuming completeness and soundness criteria are met. Because these triggers are designed to be adaptive, the SMARt model accommodates the safe, controlled expansion of an agent's operational scope over time. We conclude that formalizing failure management within the autonomy lifecycle is a crucial step toward realizing reliable and governed artificial intelligence.

Recent advances in generative models, such as diffusion models, have raised concerns related to privacy, copyright infringement, and data stewardship. To better understand and control these risks, prior work has introduced techniques and attacks that reconstruct images, or parts of images, from training data. While these results demonstrate that training data can be recovered, existing methods often rely on high computational resources, partial access to the training set, or carefully engineered prompts. In this work, we present a new attack that requires low resources, assumes little to no access to the training data, and identifies seemingly benign prompts that can lead to potentially risky image reconstruction. We further show that such reconstructions may occur unintentionally, even for users without specialized knowledge. For example, we observe that for one existing model, the prompt ``blue Unisex T-Shirt'' generates the face of a real individual. Moreover, by combining the identified vulnerabilities with real-world prompt data, we discover prompts that reproduce memorized visual elements. Our approach builds on insights from prior work and leverages domain knowledge to expose a fundamental vulnerability arising from the use of scraped e-commerce data, where templated layouts and images are closely tied to pattern-like textual prompts. The code for our attack is publicly available at https://github.com/TheSolY/lr-tmi.

The growing sophistication of synthetic image and deepfake generation models has turned source attribution and authenticity verification into a critical challenge for modern computer vision systems. Recent studies suggest that diffusion pipelines unintentionally imprint persistent statistical traces, known as signal-leaks, within their outputs, particularly in latent representations. Building on this observation, we propose Proto-LeakNet, a signal-leak-aware and interpretable attribution framework that integrates Closed-set classification with a density-based Open-set evaluation on the learned embeddings, enabling analysis of unseen generators without retraining. Acting in the latent domain of diffusion models, our method re-simulates partial forward diffusion to expose residual generator-specific cues. A temporal attention encoder aggregates multi-step latent features, while a feature-weighted prototype head structures the embedding space and enables transparent attribution. Trained solely on closed data and achieving a Macro AUC of 98.13\%, Proto-LeakNet learns a latent geometry that remains robust under post-processing, surpassing state-of-the-art methods, and achieves strong separability both between real images and known generators, and between known and unseen ones. The codebase is available at the following link: https://github.com/claudiunderthehood/Proto-LeakNet .

A goal of interpretability is to recover disentangled representations of latent concepts (features) from the activations of neural networks. The quality of features is typically evaluated in isolation, and under implicit independence assumptions that may not hold in practice. Thus, it is unclear to what extent common featurization methods such as sparse autoencoders (SAEs) and probes disentangle one concept from another. We propose a multi-concept evaluation setting using concepts including sentiment, domain, voice, and tense. We evaluate how well featurizers produce disentangled representations of each concept, observing that features are typically sensitive to only one concept, but also that concepts are distributed across many features. Then, we steer these features, measuring whether each concept is independently manipulable, and whether features interact. Even in idealized settings, steering a feature often affects many concepts, despite a near absence of interaction effects. These results suggest that correlational metrics are insufficient to establish steering selectivity, and that demonstrating that two features operate in separate spaces is insufficient to claim that they will be selective for one concept. These results underscore the importance of multi-concept evaluations in interpretability research.

Federated learning (FL) enables collaborative model training over privacy-sensitive, distributed data, but its environmental impact is difficult to compare across studies due to inconsistent measurement boundaries and heterogeneous reporting. We present a practical carbon-accounting methodology for FL CO2e tracking using NVIDIA NVFlare and CodeCarbon for explicit, phase-aware tasks (initialization, per-round training, evaluation, and idle/coordination). To capture non-compute effects, we additionally estimate communication emissions from transmitted model-update sizes under a network-configurable energy model. We validate the proposed approach on two representative workloads: CIFAR-10 image classification and retinal optic disk segmentation. In CIFAR-10, controlled client-efficiency scenarios show that system-level slowdowns and coordination effects can contribute meaningfully to carbon footprint under an otherwise fixed FL protocol, increasing total CO2e by 8.34x (medium) and 21.73x (low) relative to the high-efficiency baseline. In retinal segmentation, swapping GPU tiers (H100 vs.\ V100) yields a consistent 1.7x runtime gap (290 vs. 503 minutes) while producing non-uniform changes in total energy and CO2e across sites, underscoring the need for per-site and per-round reporting. Overall, our results support a standardized carbon accounting method that acts as a prerequisite for reproducible 'green' FL evaluation. Our code is available at https://github.com/Pediatric-Accelerated-Intelligence-Lab/carbon_footprint.

LLM-based agents are becoming increasingly capable, yet their safety lags behind. This creates a gap between what agents can do and should do. This gap widens as agents engage in multi-turn interactions and employ diverse tools, introducing new risks overlooked by existing benchmarks. To systematically scale safety testing into multi-turn, tool-realistic settings, we propose a principled taxonomy that transforms single-turn harmful tasks into multi-turn attack sequences. Using this taxonomy, we construct MT-AgentRisk (Multi-Turn Agent Risk Benchmark), the first benchmark to evaluate multi-turn tool-using agent safety. Our experiments reveal substantial safety degradation: the Attack Success Rate (ASR) increases by 16% on average across open and closed models in multi-turn settings. To close this gap, we propose ToolShield, a training-free, tool-agnostic, self-exploration defense: when encountering a new tool, the agent autonomously generates test cases, executes them to observe downstream effects, and distills safety experiences for deployment. Experiments show that ToolShield effectively reduces ASR by 30% on average in multi-turn interactions. Our code is available at https://github.com/CHATS-lab/ToolShield.

The emergence of writable, cross-session persistent memory in LLM agents introduces a qualitatively different threat landscape from conventional input-centric security concerns, characterized by three properties: persistence, statefulness, and propagation. To systematically characterize this landscape, we propose a Memory Lifecycle Framework that organizes attacks, defenses, and their cross-phase dependencies along two axes: six lifecycle phases (Write, Store, Retrieve, Execute, Share & Propagate, Forget & Rollback) and four security objectives (Integrity, Confidentiality, Availability, Governance). This analysis in turn exposes the need for formal security guarantees at the system level, motivating Verifiable Memory Governance(VMG), a framework of five architectural primitives that specifies what verifiable mechanisms a long-term-memory system must provide to maintain auditable, recoverable control over its memory state. Our analysis indicates that robust Long-Term Memory (LTM) security cannot be retrofitted at retrieval or execution time alone, but must be anchored in storage-time provenance, versioning, and policy-aware retention from the outset.

Recent work on text diffusion models offers a promising alternative to autoregressive generation, but controlling their safety remains underexplored. Existing safety approaches are geared toward autoregressive models and typically rely on post-hoc filtering or inference-time interventions. These are inadequate for effectively addressing safety risks in text diffusion models. We propose the Safety-Aware Denoiser (SAD), a safety-guidance framework in text diffusion models. The SAD modifies the iterative denoising process such that the text sample at the final denoising step is steered toward provably safe regions of the text space. This inference-time method can integrate safety constraints into the denoiser, avoiding computationally expensive retraining of the underlying diffusion model and enabling flexible, lightweight safety guidance. We evaluate the safety of the generated text using the SAD, with respect to hazard taxonomy, memorization, and jailbreak. Experimental results show that SAD substantially reduces unsafe generations while preserving generation quality, diversity, and fluency, outperforming existing methods. These results demonstrate that our safety guidance during denoising provides an effective and scalable mechanism for enforcing safety in text diffusion models.

Mechanistic interpretability often studies Transformer behavior by intervening on internal activations through activation patching, causal tracing, path patching, and steering directions. This paper develops Transformer Field Theory: a response-theoretic framework in which the residual stream of a fixed forward pass is treated as a Transformer field over layer depth and token position. In this formulation, patching becomes a localized source insertion into the Transformer field, first-order sensitivity fields predict patch effects, Green functions describe downstream propagation, and patch selection is posed as an adjoint inverse problem. Empirically, we test the theory's forward response objects in GPT-2-style autoregressive Transformers. Localized Transformer-field interventions exhibit a bounded local linear regime; first-order sensitivities predict patch effects across layer-token sites; localized sources generate structured anisotropic Transformer-field propagation; high-sensitivity sites and sliced Green operators provide reduced response descriptions; and prompt-induced Transformer-field displacements partially transfer answer behavior. These results establish sensitivities, Transformer-field responses, and sliced Green operators as practical objects for organizing patching experiments, while providing the forward mathematical basis for patch-site inference and cross-scale response transfer.

Two-sample testing is a fundamental tool for detecting distributional differences across scientific domains, but classical tests (including kernel-based tests) can be ineffective on high-dimensional structured data such as images. Recent deep two-sample tests improve sensitivity in these settings by learning informative representations, yet they provide limited insight into which data features drive rejection of the null hypothesis $H_0$. To address this issue, we propose a counterfactual explanation framework for deep two-sample testing that generates sample-level edits moving observations from a source group toward a target group while explicitly reducing the discrepancy measured by the test. Our method combines a diffusion autoencoder with a pretrained deep two-sample test model and optimizes a maximum mean discrepancy (MMD) objective in the test model's representation space to produce plausible counterfactuals. We quantify distribution-level effects through changes in the test statistic and the resulting two-sample p-values. We evaluate the method on synthetic 2D shape datasets and two MRI cohorts. Across both settings, the counterfactual transformations consistently increase p-values relative to the original samples, indicating that the edited source set becomes statistically closer to the target distribution under the test. We measure minimality using LPIPS to ensure the counterfactuals remain close to the original samples. The resulting edits provide interpretable evidence of the features associated with the detected group differences. On MRI, the localized changes are consistent with known anatomical differences between cohorts.

Reinforcement learning from human feedback (RLHF) is bottlenecked by reward hacking, where the policy exploits errors in a proxy reward model (RM) and produces high RM scores without genuine quality gains. A natural mitigation is pessimism: lowering rewards in regions where the RM is uncertain. However, standard scalar RMs provide no principled notion of uncertainty. We argue that the right object is a distributional reward model $p(r\mid x,y)$. Under either a Bayesian inference or a KL-distributionally robust optimization (KL-DRO) lens, the KL-regularized RLHF objective admits a closed-form effective reward $\tilde r(x,y) = \pmβ\log\mathbb{E}_p[e^{\pm r/β}]$. The pessimistic branch unifies the prior heuristics for RM ensemble aggregation: mean aggregation, worst-case optimization (WCO), and uncertainty-weighted optimization (UWO) all emerge as limits or truncations of this single expression. This also clarifies the implicit assumptions of each existing rule.

Large language models deployed as agents over large tool catalogs face a critical tool-retrieval bottleneck. As embedding-based retrieval approaches rely on compact encoders that may under-capture specialized tool semantics, parametric tool retrieval addresses this by encoding each tool as a virtual token appended to the LLM vocabulary, fine-tuned in two stages (memorization then retrieval SFT) to use the LLM as a retriever, achieving strong performance on standard ToolBench retrieval benchmarks. Yet these benchmarks use verbose, fully-specified queries, and their evaluation applies constrained decoding that restricts outputs to valid token paths, neither reveals whether the model actually understands its tools. We introduce \textbf{ToolSense}, an open-source LLM-powered diagnostic framework that takes any tool catalog as input and automatically generates three benchmarks: a Realistic Retrieval Benchmark (RRB) with queries at three ambiguity tiers, an MCQ probing benchmark, and a QA probing benchmark. Applying ToolSense to ToolBench (~47k tools) and evaluating five parametric model training configurations reveals a knowledge-retrieval dissociation: on RRB queries, several configurations collapse by ~50-64 percentage points compared to fully-specified ToolBench benchmarks, falling below the embedding-model baseline. Additionally, despite strong retrieval performance, some models score near-random on factual probes, suggesting a knowledge-retrieval dissociation. We open-source the ToolSense framework and the ToolBench diagnostic benchmarks at https://github.com/SAP/toolsense.

Anticipating LLM behavioral tendencies from low-cost psychometric probes is critical for safe deployment, but only if self-reports (SR) reliably predict behavior. Recent work documented substantial SR-behavior dissociation in LLMs, but relied on broad personality traits (Big 5) that predict specific behaviors weakly, even in humans. Furthermore, the isolation of conversational sessions combined with weak context matching left open whether LLMs truly lack coherence or whether the conditions needed to detect such coherence were not met. We contrast Big 5 with the Theory of Planned Behavior (TPB), which measures intention targeted to a specific behavior and predicts human behavior substantially better than broad traits. We run experiments across four behavioral tasks and 11 frontier LLMs, while also varying session context and identity induction. We find that SR-behavior coherence exists but is selective. 1) Within a shared conversation, the Theory of Planned Behavior reaches human-level coherence; Big 5 does not. 2) Across separate conversations, coherence survives only for behaviors anchored outside the immediate prompt, such as implicit bias shaped by training, and collapses when behavior is strongly primed by context, as with sycophancy. 3) Persona prompting makes self-reports more consistent across conversations, but does not bring behavior into alignment. These findings suggest that coarse personality frameworks, such as Big 5 may not be the best tools for testing deployment behavior. More task- and behavior-specific instruments are needed, and even these must be evaluated across tasks and contexts.

AI agents are increasingly being developed to accelerate scientific discovery, yet their practical capabilities in real research settings remain poorly understood. Existing benchmarks for AI agents rarely capture the complexity, heterogeneity, and extended reasoning required by scientific work, whereas benchmarks for scientific tasks often reduce research to static, direct problems and provide limited support for interactive evaluation. Here, we introduce SciAgentArena, a systematic benchmark for evaluating AI agents in real-world scientific research scenarios drawn from emerging needs across multiple domains. SciAgentArena comprises approximately 200 tasks with stepwise verification and an interactive, agent-agnostic environment for assessing diverse AI agents. Using this benchmark, we find that current agents can contribute effectively to well-specified data-analysis workflows, particularly when the task structure and evaluation criteria are clear. However, their performance remains uneven across scientific contexts: agents struggle to generate genuinely novel insights, sustain self-directed exploration, and formulate robust solutions for open-ended research questions. We further characterize common failure modes across agents and identify opportunities for improving their reliability, autonomy, and scientific reasoning. Together, SciAgentArena provides a practical framework for measuring progress in AI agents for science and for guiding the design of future agents capable of addressing complex scientific challenges. Full codes, tasks, and datasets can be accessed via this link: https://sciagentarena.github.io/.