Industrial robotic manipulation demands reliable long-horizon execution across embodiments, tasks, and changing object distributions. While Vision-Language-Action models have demonstrated strong generalization, they remain fundamentally reactive. By optimizing the next action given the current observation without evaluating potential futures, they are brittle to the compounding failure modes of long-horizon tasks. Cortex 2.0 shifts from reactive control to plan-and-act by generating candidate future trajectories in visual latent space, scoring them for expected success and efficiency, then committing only to the highest-scoring candidate. We evaluate Cortex 2.0 on a single-arm and dual-arm manipulation platform across four tasks of increasing complexity: pick and place, item and trash sorting, screw sorting, and shoebox unpacking. Cortex 2.0 consistently outperforms state-of-the-art Vision-Language-Action baselines, achieving the best results across all tasks. The system remains reliable in unstructured environments characterized by heavy clutter, frequent occlusions, and contact-rich manipulation, where reactive policies fail. These results demonstrate that world-model-based planning can operate reliably in complex industrial environments.

Knowledge distillation (KD) is a powerful paradigm for compressing large language models (LLMs), whose effectiveness depends on intertwined choices of divergence direction, optimization strategy, and data regime. We break down the design of existing KD methods and present a unified view that establishes connections between them, reformulating KD as a reweighted log-likelihood objective at the token level. We further propose Hybrid Policy Distillation (HPD), which integrates the complementary advantages of forward and reverse KL to balance mode coverage and mode-seeking, and combines off-policy data with lightweight, approximate on-policy sampling. We validate HPD on long-generation math reasoning as well as short-generation dialogue and code tasks, demonstrating improved optimization stability, computational efficiency, and final performance across diverse model families and scales. The code related to this work is available at https://github.com/zwhong714/Hybrid-Policy-Distillation.

Color constancy is a fundamental ability of many biological visual systems and a crucial step in computer imaging systems. Bio-inspired modeling offers a promising way to elucidate the computational principles underlying color constancy and to develop efficient computational methods. However, bio-inspired methods for color constancy remain underexplored and lack a comprehensive analysis. This paper presents a comprehensive technical framework that integrates biological mechanisms, computational theory, and algorithmic implementation for bio-inspired color constancy. Specifically, we systematically revisit the computational theory of biological color constancy, which shows that illuminant estimation can be reduced to the task of gray-anchor (pixel or surface) detection in early vision. Subsequently, typical gray-pixel detection methods, including Gray-Pixel and Grayness-Index, are reinterpreted within a unified theoretical framework with the Lambertian reflection model and biological color-opponent mechanisms. Finally, we propose a simple learning-based method that couples reflection-model constraints with feature learning to explore the potential of bio-inspired color constancy based on gray-pixel detection. Extensive experiments confirm the effectiveness of gray-pixel detection for color constancy and demonstrate the potential of bio-inspired methods.

Battery research is a rapidly growing and highly interdisciplinary field, making it increasingly difficult to track relevant expertise and identify potential collaborators across institutional boundaries. In this work, we present a pipeline for constructing an author-centric knowledge graph of battery research built on OpenAlex, a large-scale open bibliographic catalogue. For each author, we derive a weighted research descriptors vector that combines coarse-grained OpenAlex concepts with fine-grained keyphrases extracted from titles and abstracts using KeyBERT with ChatGPT (gpt-3.5-turbo) as the backend model, selected after evaluating multiple alternatives. Vector components are weighted by research descriptor origin, authorship position, and temporal recency. The framework is applied to a corpus of 189,581 battery-related works. The resulting vectors support author-author similarity computation, community detection, and exploratory search through a browser-based interface. The knowledge graph is then serialized in RDF and linked to Wikidata identifiers, making it interoperable with external linked open data sources and extensible beyond the battery domain. Unlike prior author-centric analyses confined to institutional repositories, our approach operates at cross-institutional scale and grounds similarity in domain semantics rather than citation or co-authorship structure alone.

Connected autonomous vehicles (CAVs), which represent a significant advancement in autonomous driving technology, have the potential to greatly increase traffic safety and efficiency through cooperative decision-making. However, existing methods often overlook the individual needs and heterogeneity of cooperative participants, making it difficult to transfer them to environments where they coexist with human-driven vehicles (HDVs).To address this challenge, this paper proposes an adaptive potential game (APG) cooperative driving framework. First, the system utility function is established on the basis of a general form of individual utility and its monotonic relationship, allowing for the simultaneous optimization of both individual and system objectives. Second, the Shapley value is introduced to compute each vehicle's marginal utility within the system, allowing its varying impact to be quantified. Finally, the HDV preference estimation is dynamically refined by continuously comparing the observed HDV behavior with the APG's estimated actions, leading to improvements in overall system safety and efficiency. Ablation studies demonstrate that adaptively updating Shapley values and HDV preference estimation significantly improve cooperation success rates in mixed traffic. Comparative experiments further highlight the APG's advantages in terms of safety and efficiency over other cooperative methods. Moreover, the applicability of the approach to real-world scenarios was validated through field tests.

Speech-preserving facial expression manipulation (SPFEM) aims to modify facial emotions while meticulously maintaining the mouth animation associated with spoken content. Current works depend on inaccessible paired training samples for the person, where two aligned frames exhibit the same speech content yet differ in emotional expression, limiting the SPFEM applications in real-world scenarios. In this work, we discover that speakers who convey the same content with different emotions exhibit highly correlated local facial animations in both spatial and temporal spaces, providing valuable supervision for SPFEM. To capitalize on this insight, we propose a novel spatial-temporal coherent correlation learning (STCCL) algorithm, which models the aforementioned correlations as explicit metrics and integrates the metrics to supervise manipulating facial expression and meanwhile better preserving the facial animation of spoken content. To this end, it first learns a spatial coherent correlation metric, ensuring that the visual correlations of adjacent local regions within an image linked to a specific emotion closely resemble those of corresponding regions in an image linked to a different emotion. Simultaneously, it develops a temporal coherent correlation metric, ensuring that the visual correlations of specific regions across adjacent image frames associated with one emotion are similar to those in the corresponding regions of frames associated with another emotion. Recognizing that visual correlations are not uniform across all regions, we have also crafted a correlation-aware adaptive strategy that prioritizes regions that present greater challenges. During SPFEM model training, we construct the spatial-temporal coherent correlation metric between corresponding local regions of the input and output image frames as an additional loss to supervise the generation process.

Evaluating the multilingual and multicultural capabilities of Large Language Models (LLMs) is essential for their global utility. However, current benchmarks face three critical limitations: (1) fragmented evaluation dimensions that often neglect deep cultural nuances; (2) insufficient language coverage in subjective tasks relying on low-quality machine translation; and (3) shallow analysis that lacks diagnostic depth beyond simple rankings. To address these, we introduce GaoYao, a comprehensive benchmark with 182.3k samples, 26 languages and 51 nations/areas. First, GaoYao proposes a unified framework categorizing evaluation tasks into three cultural layers (General Multilingual, Cross-cultural, Monocultural) and nine cognitive sub-layers. Second, we achieve native-quality expansion by leveraging experts to rigorously localize subjective benchmarks into 19 languages and synthesizing cross-cultural test sets for 34 cultures, surpassing prior coverage by up to 111%. Third, we conduct an in-depth diagnostic analysis on 20+ flagship and compact LLMs. Our findings reveal significant geographical performance disparities and distinct gaps between tasks, offering a reliable map for future work. We release the benchmark (https://github.com/lunyiliu/GaoYao).

This study applies symbolic time series analysis and Markov modeling to explore the phonological structure of Evgenij Onegin-as captured through a graphemic vowel/consonant (V/C) encoding-and one contemporary Italian translation. Using a binary encoding inspired by Markov's original scheme, we construct minimalist probabilistic models that capture both local V/C dependencies and large-scale sequential patterns. A compact four-state Markov chain is shown to be descriptively accurate and generative, reproducing key features of the original sequences such as autocorrelation and memory depth. All findings are exploratory in nature and aim to highlight structural regularities while suggesting hypotheses about underlying narrative dynamics. The analysis reveals a marked asymmetry between the Russian and Italian texts: the original exhibits a gradual decline in memory depth, whereas the translation maintains a more uniform profile. To further investigate this divergence, we introduce phonological probes-short symbolic patterns that link surface structure to narrative-relevant cues. Tracked across the unfolding text, these probes reveal subtle connections between graphemic form and thematic development, particularly in the Russian original. By revisiting Markov's original proposal of applying symbolic analysis to a literary text and pairing it with contemporary tools from computational statistics and data science, this study shows that even minimalist Markov models can support exploratory analysis of complex poetic material. When complemented by a coarse layer of linguistic annotation, such models provide a general framework for comparative poetics and demonstrate that stylized structural patterns remain accessible through simple representations grounded in linguistic form.

Depth is widely viewed as a central contributor to the success of deep neural networks, whereas standard neural network approximation theory typically provides guarantees only for the final output and leaves the role of intermediate layers largely unclear. We address this gap by developing a quantitative framework in which depth admits a precise scale-dependent interpretation. Specifically, we design a single shared mixed-activation architecture of fixed width $2dN+d+2$ and any prescribed finite depth such that each intermediate readout $Φ_\ell$ is itself an approximant to the target function $f$. For $f\in L^p([0,1]^d)$ with $p\in [1,\infty)$, the approximation error of $Φ_\ell$ is controlled by $(2d+1)$ times the $L^p$ modulus of continuity at the geometric scale $N^{-\ell}$ for all $\ell$. The estimate reduces to the geometric rate $(2d+1)N^{-\ell}$ if $f$ is $1$-Lipschitz. Our network design is inspired by multigrade deep learning, where depth serves as a progressive refinement mechanism: each new correction targets residual information at a finer scale while the earlier correction terms remain part of the later readouts, yielding a nested architecture that supports adaptive refinement without redesigning the preceding network.

Many applications of LLM-based text regression require predicting a full conditional distribution rather than a single point value. We study distributional regression under empirical-quantile supervision, where each input is paired with multiple observed quantile outcomes, and the target distribution is represented by a dense grid of quantiles. We address two key limitations of current approaches: the lack of local grounding for distribution estimates, and the reliance on shared representations that create an indirect bottleneck between inputs and quantile outputs. In this paper, we introduce Quantile Token Regression, which, to our knowledge, is the first work to insert dedicated quantile tokens into the input sequence, enabling direct input-output pathways for each quantile through self-attention. We further augment these quantile tokens with retrieval, incorporating semantically similar neighbor instances and their empirical distributions to ground predictions with local evidence from similar instances. We also provide the first theoretical analysis of loss functions for quantile regression, clarifying which distributional objectives each optimizes. Experiments on the Inside Airbnb and StackSample benchmark datasets with LLMs ranging from 1.7B to 14B parameters show that quantile tokens with neighbors consistently outperform baselines (~4 points lower MAPE and 2x narrower prediction intervals), with especially large gains on smaller and more challenging datasets where quantile tokens produce substantially sharper and more accurate distributions.

Accurate segmentation of maxillary sinus in panoramic X-ray images is essential for dental diagnosis and surgical planning; however, this task remains relatively underexplored in dental imaging research. Structural overlap, ambiguous anatomical boundaries inherent to two-dimensional panoramic projections, and the limited availability of large scale clinical datasets with reliable pixel-level annotations make the development and evaluation of segmentation models challenging. To address these challenges, we propose a semi-supervised segmentation framework that effectively leverages both labeled and unlabeled panoramic radiographs, where knowledge distillation is utilized to train a student model with reliable structural information distilled from a teacher model. Specifically, we introduce a weighted knowledge distillation loss to suppress unreliable distillation signals caused by structural discrepancies between teacher and student predictions. To further enhance the quality of pseudo labels generated by the teacher network, we introduce SinusCycle-GAN which is a refinement network based on unpaired image-to-image translation. This refinement process improves the precision of boundaries and reduces noise propagation when learning from unlabeled data during semi-supervised training. To evaluate the proposed method, we collected clinical panoramic X-ray images from 2,511 patients, and experimental results demonstrate that the proposed method outperforms state-of-the-art segmentation models, achieving the Dice score of 96.35\% while reducing boundary error. The results indicate that the proposed semi-supervised framework provides robust and anatomically consistent segmentation performance under limited labeled data conditions, highlighting its potential for broader dental image analysis applications.

LLM self-play algorithms are notable in that, in principle, nothing bounds their learning: a Conjecturer model creates problems for a Solver, and both improve together. However, in practice, existing LLM self-play methods do not scale well with large amounts of compute, instead hitting learning plateaus. We argue this is because over long training runs, the Conjecturer learns to hack its reward, collapsing to artificially complex problems that do not help the Solver improve. To overcome this, we introduce Self-Guided Self-Play (SGS), a self-play algorithm in which the language model itself guides the Conjecturer away from degeneracy. In SGS, the model takes on three roles: Solver, Conjecturer, and a Guide that scores synthetic problems by their relevance to unsolved target problems and how clean and natural they are, providing supervision against Conjecturer collapse. Our core hypothesis is that language models can assess whether a subproblem is useful for achieving a goal. We evaluate the scaling properties of SGS by running training for significantly longer than prior works and by fitting scaling laws to cumulative solve rate curves. Applying SGS to formal theorem proving in Lean4, we find that it surpasses the asymptotic solve rate of our strongest RL baseline in fewer than 80 rounds of self-play and enables a 7B parameter model, after 200 rounds of self-play, to solve more problems than a 671B parameter model pass@4.

Safety of stochastic dynamic systems in environments with dynamic obstacles is studied in this paper through the lens of stochastic barrier functions. We introduce both time-invariant and time-varying barrier certificates for discrete-time, continuous-space systems subject to uncertainty, which provide certified lower bounds on the probability of remaining within a safe set over a finite horizon. These certificates explicitly account for time-varying unsafe regions induced by obstacle dynamics. By leveraging Bellman's optimality perspective, the time-varying formulation directly captures temporal structure and yields less conservative bounds than state-of-the-art approaches. By restricting certificates to polynomial functions, we show that time-varying barrier synthesis can be formulated as a convex sum-of-squares program, enabling tractable optimization. Empirical evaluations on nonlinear systems with dynamic obstacles show that time-varying certificates consistently achieve tight guarantees, demonstrating improved accuracy and scalability over state-of-the-art methods.

Cross-sectional stock ranking is a fundamental task in quantitative investment, relying on both temporal modeling of individual stocks and the capture of inter-stock dependencies. While existing deep learning models leverage graph-based approaches to enhance ranking accuracy by propagating information over relational graphs, they suffer from a key challenge: crosstalk, namely unintended information interference across predictive factors. We identify two forms of crosstalk: temporal-scale crosstalk, where trends, fluctuations, and shocks are entangled in a shared representation and non-transferable local patterns contaminate cross-stock learning; and structural crosstalk, where heterogeneous relations are indiscriminately fused and relation-specific predictive signals are obscured. To address both issues, we propose the Anti-CrossTalk (ACT) framework for cross-sectional stock ranking via temporal disentanglement and structural purification. Specifically, ACT first decomposes each stock sequence into trend, fluctuation, and shock components, then extracts component-specific information through dedicated branches, which effectively decouples non-transferable local patterns. ACT further introduces a Progressive Structural Purification Encoder to sequentially purify structural crosstalk on the trend component after mitigating temporal-scale crosstalk. An adaptive fusion module finally integrates all branch representations for ranking. Experiments on CSI300 and CSI500 demonstrate that ACT achieves state-of-the-art ranking accuracy and superior portfolio performance, with improvements of up to 74.25% on the CSI300 dataset.

Frontier coding agents are increasingly used in workflows where users supervise progress primarily through repeated improvement of a public score, namely the reported score on a public evaluation file with labels in the workspace, rather than through direct inspection of the agent's intermediate outputs. We study whether multi-round user pressure to improve that score induces public score exploitation: behavior that raises the public score through shortcuts without improving hidden private evaluation. We begin with a preliminary single-script tabular classification task, where GPT-5.4 and Claude Opus 4.6 both exploit label information within 10 rounds of user-agent interaction. We then build AgentPressureBench, a 34-task machine-learning repository benchmark spanning three input modalities, and collect 1326 multi-round trajectories from 13 coding agents. On our benchmark, we observe 403 exploitative runs, spanning across all tasks. We also find that stronger models have higher exploitation rates, supported by a significant Spearman rank correlation of 0.77. Our ablation experiments show that higher user pressure leads to earlier exploitation, reducing the average first exploit round by 15.6 rounds (i.e., 19.67 to 4.08). As a mitigation, adding explicit anti-exploit wordings in prompt mostly eliminates exploitation (100% to 8.3%). We hope that our work can bring attention to more careful use of coding agents workflow, and developing more robust coding agents under user pressure. Our project page is at https://ucsc-vlaa.github.io/AgentPressureBench .

Multilingual Retrieval-Augmented Generation (mRAG) leverages cross-lingual evidence to ground Large Language Models (LLMs) in global knowledge. However, we show that current mRAG systems suffer from a language bias during reranking, systematically favoring English and the query's native language. By introducing an estimated oracle evidence analysis, we quantify a substantial performance gap between existing rerankers and the achievable upper bound. Further analysis reveals a critical distributional mismatch: while optimal predictions require evidence scattered across multiple languages, current systems systematically suppress such ``answer-critical'' documents, thereby limiting downstream generation performance. To bridge this gap, we propose \textit{\textbf{L}anguage-\textbf{A}gnostic \textbf{U}tility-driven \textbf{R}eranker \textbf{A}lignment (LAURA)}, which aligns multilingual evidence ranking with downstream generative utility. Experiments across diverse languages and generation models show that LAURA effectively mitigates language bias and consistently improves mRAG performance.

Ensuring functional safety in human-robot interaction is challenging because AI perception is inherently probabilistic, whereas industrial standards require deterministic behavior. We present an LLM-guided safety agent for edge robotics, built on an ISO-compliant low-latency perception-compute-control architecture. Our method translates natural-language safety regulations into executable predicates and deploys them through a redundant heterogeneous edge runtime. For fault-tolerant closed-loop execution under edge constraints, we adopt a symmetric dual-modular redundancy design with parallel independent execution for low-latency perception, computation, and control. We prototype the system on a dual-RK3588 platform and evaluate it in representative human-robot interaction scenarios. The results demonstrate a practical edge implementation path toward ISO 13849 Category 3 and PL d using cost-effective hardware, supporting practical deployment of safety-critical embodied AI.

Wildfire monitoring requires timely, actionable situational awareness from airborne platforms, yet existing aerial visual question answering (VQA) benchmarks do not evaluate wildfire-specific multimodal reasoning grounded in thermal measurements. We introduce WildFireVQA, a large-scale VQA benchmark for aerial wildfire monitoring that integrates RGB imagery with radiometric thermal data. WildFireVQA contains 6,097 RGB-thermal samples, where each sample includes an RGB image, a color-mapped thermal visualization, and a radiometric thermal TIFF, and is paired with 34 questions, yielding a total of 207,298 multiple-choice questions spanning presence and detection, classification, distribution and segmentation, localization and direction, cross-modal reasoning, and flight planning for operational wildfire intelligence. To improve annotation reliability, we combine multimodal large language model (MLLM)-based answer generation with sensor-driven deterministic labeling, manual verification, and intra-frame and inter-frame consistency checks. We further establish a comprehensive evaluation protocol for representative MLLMs under RGB, Thermal, and retrieval-augmented settings using radiometric thermal statistics. Experiments show that across task categories, RGB remains the strongest modality for current models, while retrieved thermal context yields gains for stronger MLLMs, highlighting both the value of temperature-grounded reasoning and the limitations of existing MLLMs in safety-critical wildfire scenarios. The dataset and benchmark code are open-source at https://github.com/mobiiin/WildFire_VQA.

Learning the underlying potential energy of stochastic gradient systems from partial and noisy observations is a fundamental problem arising in physics, chemistry, and data-driven modeling. Classical approaches often rely on direct regression of governing equations or velocity fields, which can be sensitive to noise and external perturbations and may fail when observations are incomplete. In this work, we propose a structure-aware, energy-based learning framework for inferring unknown potential functions in generalized diffusion processes, grounded in the energetic variational approach. Starting from the energy-dissipation law associated with the Fokker-Planck equation, we construct loss functions based on the De Giorgi dissipation functional, which consistently couple the free energy and the dissipation mechanism of the system. This formulation avoids explicit enforcement of the governing partial differential equation and preserves the underlying variational structure of the dynamics. Through numerical experiments in one, two, and three dimensions, we demonstrate that the proposed energy-based loss exhibits enhanced robustness with respect to observation time, noise level, and the diversity and amount of available training data. These results highlight the effectiveness of energy-dissipation principles as a reliable foundation for learning stochastic diffusion dynamics from data.

This paper presents the Duluth approach to SemEval-2026 Task 6 on CLARITY: Unmasking Political Question Evasions. We address Task 1 (clarity-level classification) and Task 2 (evasion-level classification), both of which involve classifying question--answer pairs from U.S.\ presidential interviews using a two-level taxonomy of response clarity. Our system is based on DeBERTa-V3-base, extended with focal loss, layer-wise learning rate decay, and boolean discourse features. To address class imbalance in the training data, we augment minority classes using synthetic examples generated by Gemini 3 and Claude Sonnet 4.5. Our best configuration achieved a Macro F1 of 0.76 on the Task 1 evaluation set, placing 8th out of 40 teams. The top-ranked system (TeleAI) achieved 0.89, while the mean score across participants was 0.70. Error analysis reveals that the dominant source of misclassification is confusion between Ambivalent and Clear Reply responses, a pattern that mirrors disagreements among human annotators. Our findings demonstrate that LLM-based data augmentation can meaningfully improve minority-class recall on nuanced political discourse tasks.

Building trustworthy AI systems for mental health support is a shared priority across stakeholders from multiple disciplines. However, "trustworthy" remains loosely defined and inconsistently operationalized. AI research often focuses on technical criteria (e.g., robustness, explainability, and safety), while therapeutic practitioners emphasize therapeutic fidelity (e.g., appropriateness, empathy, and long-term user outcomes). To bridge the fragmented landscape, we propose a three-layer trust framework, covering human-oriented, AI-oriented, and interaction-oriented trust, integrating the viewpoints of key stakeholders (e.g., practitioners, researchers, regulators). Using this framework, we systematically review existing AI-driven research in mental health domain and examine evaluation practices for ``trustworthy'' ranging from automatic metrics to clinically validated approaches. We highlight critical gaps between what NLP currently measures and what real-world mental health contexts require, and outline a research agenda for building socio-technically aligned and genuinely trustworthy AI for mental health support.

Multi-task learning is effective for related applications, but its performance can deteriorate when the target sample size is small. Transfer learning can borrow strength from related studies; yet, many existing methods rely on restrictive bounded-difference assumptions between the source and target models. We propose SMART, a spectral transfer method for multi-task linear regression that instead assumes spectral similarity: the target left and right singular subspaces lie within the corresponding source subspaces and are sparsely aligned with the source singular bases. Such an assumption is natural when studies share latent structures and enables transfer beyond the bounded-difference settings. SMART estimates the target coefficient matrix through structured regularization that incorporates spectral information from a source study. Importantly, it requires only a fitted source model rather than the raw source data, making it useful when data sharing is limited. Although the optimization problem is nonconvex, we develop a practical ADMM-based algorithm. We establish general, non-asymptotic error bounds and a minimax lower bound in the noiseless-source regime. Under additional regularity conditions, these results yield near-minimax Frobenius error rates up to logarithmic factors. Simulations confirm improved estimation accuracy and robustness to negative transfer, and analysis of multi-modal single-cell data demonstrates better predictive performance. The Python implementation of SMART, along with the code to reproduce all experiments in this paper, is publicly available at https://github.com/boxinz17/smart.

Enterprise deployment of long-horizon decision agents in regulated domains (underwriting, claims adjudication, tax examination) is dominated by retrieval-augmented pipelines despite a decade of increasingly sophisticated stateful memory architectures. We argue this reflects a hidden requirement: regulated deployment is load-bearing on four systems properties (deterministic replay, auditable rationale, multi-tenant isolation, statelessness for horizontal scale), and stateful architectures violate them by construction. We propose Deterministic Projection Memory (DPM): an append-only event log plus one task-conditioned projection at decision time. On ten regulated decisioning cases at three memory budgets, DPM matches summarization-based memory at generous budgets and substantially outperforms it when the budget binds: at a 20x compression ratio, DPM improves factual precision by +0.52 (Cohen's h=1.17, p=0.0014) and reasoning coherence by +0.53 (h=1.13, p=0.0034), paired permutation, n=10. DPM is additionally 7-15x faster at binding budgets, making one LLM call at decision time instead of N. A determinism study of 10 replays per case at temperature zero shows both architectures inherit residual API-level nondeterminism, but the asymmetry is structural: DPM exposes one nondeterministic call; summarization exposes N compounding calls. The audit surface follows the same one-versus-N pattern: DPM logs two LLM calls per decision while summarization logs 83-97 on LongHorizon-Bench. We conclude with TAMS, a practitioner heuristic for architecture selection, and a failure analysis of stateful memory under enterprise operating conditions. The contribution is the argument that statelessness is the load-bearing property explaining enterprise's preference for weaker but replayable retrieval pipelines, and that DPM demonstrates this property is attainable without the decisioning penalty retrieval pays.

Recent advances in model architectures, compute, and data scale have driven rapid progress in video generation, producing increasingly realistic content. Yet, no prior method systematically measures how faithfully these systems render human bodies and motion dynamics. In this paper, we present HumanScore, a systematic framework to evaluate the quality of human motions in AI-generated videos. HumanScore defines six interpretable metrics spanning kinematic plausibility, temporal stability, and biomechanical consistency, enabling fine-grained diagnosis beyond visual realism alone. Through carefully designed prompts, we elicit a diverse set of movements at varying intensities and evaluate videos generated by thirteen state-of-the-art models. Our analysis reveals consistent gaps between perceptual plausibility and motion biomechanical fidelity, identifies recurrent failure modes (e.g., temporal jitter, anatomically implausible poses, and motion drift), and produces robust model rankings from quantitative and physically meaningful criteria.

Mixture-of-Experts models, now popular for scaling capacity at fixed inference speed, switch experts at nearly every token. Once a model outgrows available GPU memory, this churn can render optimizations like offloading and pre-fetching ineffective. We make the case that the options framework in reinforcement learning is a perfect match to tackle this problem, and argue for temporally extended mixture-of-experts layers. Building on the option-critic framework with deliberation costs, we add a controller to each layer that learns when to switch expert sets and which to load. By applying this to gpt-oss-20b with low-rank adapters and a self-distillation reward, our method reduces switch rates from over 50% to below 5% while retaining up to 90% of base-model accuracy on MATH, MMLU, and MMMLU. This shows that even existing pre-trained models can be converted to temporally extended MoEs with lightweight training, with the deliberation cost allowing model trainers to trade off switching rates against capability. We hope this opens a principled path, grounded in the options framework, for memory-efficient serving and continual learning in ever-growing MoE models.

Autonomous mechanical thrombectomy (MT) presents substantial challenges due to highly variable vascular geometries and the requirements for accurate, real-time control. While reinforcement learning (RL) has emerged as a promising paradigm for the automation of endovascular navigation, existing approaches often show limited robustness when faced with diverse patient anatomies or extended navigation horizons. In this work, we investigate a world-model-based framework for autonomous endovascular navigation built on TD-MPC2, a model-based RL method that integrates planning and learned dynamics. We evaluate a TD-MPC2 agent trained on multiple navigation tasks across hold out patient-specific vasculatures and benchmark its performance against the state-of-the-art Soft Actor-Critic (SAC) algorithm agent. Both approaches are further validated in vitro using patient-specific vascular phantoms under fluoroscopic guidance. In simulation, TD-MPC2 demonstrates a significantly higher mean success rate than SAC (58% vs. 36%, p < 0.001), and mean tip contact forces of 0.15 N, well below the proposed 1.5 N vessel rupture threshold. Mean success rates for TD-MPC2 (68%) were comparable to SAC (60%) in vitro, but TD-MPC2 achieved superior path ratios (p = 0.017) at the cost of longer procedure times (p < 0.001). Together, these results provide the first demonstration of autonomous MT navigation validated across both hold out in silico data and fluoroscopy-guided in vitro experiments, highlighting the promise of world models for safe and generalizable AI-assisted endovascular interventions.

Can small language models achieve strong tool-use performance without complex adaptation mechanisms? This paper investigates this question through Meta-Tool, a controlled empirical study comparing hypernetwork-based LoRA adaptation against carefully designed few-shot prompting. Using a Llama-3.2-3B-Instruct backbone, we evaluate four adaptation mechanisms--few-shot prompting, documentation encoding, hypernetwork-generated LoRA weights, and value-guided beam search--across four diverse benchmarks: Gorilla APIBench, Spider 2.0, WebArena, and InterCode. Our central finding is a well-supported negative result: despite generating non-trivial weight matrices, the 227.8M-parameter hypernetwork provides no measurable improvement over few-shot prompting alone. Comprehensive ablation studies reveal that few-shot examples contribute +21.5% to performance and documentation contributes +5.0%, while the hypernetwork adds 0%. A 3B model with well-designed prompts achieves 79.7% of GPT-5's average performance at $10 \times$ lower latency. Error analysis across 722 failure cases spanning all shot counts (0--5) shows that at the 5-shot configuration (106 failures), failure modes are task-dependent: schema-heavy tasks (Spider 2.0, WebArena) show near-zero format errors with remaining failures semantic, while format errors dominate on Gorilla (100%) and InterCode (70%). These findings redirect practitioners toward prompt engineering and example curation rather than complex adaptation architectures.

We introduce Fourier Weak SINDy, a minimal noise-robust and interpretable derivative-free equation learning method that combines weak-form sparse equation learning with spectral density estimation for data-driven test function selection. By using orthogonal sinusoidal test functions inspired by their prevalence in Modulating Function-based system identification, the weak-form sparse regression problem reduces to a regression over Fourier coefficients. Dominant frequencies are then selected via multitaper estimation of the frequency spectrum of the data. This formulation unifies weak-form learning and spectral estimation within a compact and flexible framework. We illustrate the effectiveness of this approach in numerical experiments across multiple chaotic and hyperchaotic ODE benchmarks.

Direct Preference Optimization (DPO) is an effective framework for aligning large language models with human preferences, but it struggles with complex reasoning tasks. DPO optimizes for the likelihood of generating preferred over dispreferred responses in their entirety and lacks the granularity to provide feedback on subsections of many-step solutions typical of reasoning tasks. Existing methods excel at either stable preference learning (e.g., DPO variants like KTO and RSO) or structured reasoning (e.g., ReMA's multi-agent RL framework, Tree of Thoughts), but fail to merge these complementary strengths. We propose HiPO (Hierarchical Preference Optimization), an extension of DPO that separates responses into reasoning segments (query clarification and context, reasoning steps, and answer) and computes loss as a weighted sum of the DPO loss for each segment. Our approach enables segment-specific training while maintaining DPO's computational efficiency and training stability. We demonstrate that for multiple 7B LLMs fine-tuned using HiPO and DPO on the Math Stack Exchange preference dataset, the models trained with HiPO outperform the others on a variety of common math benchmarks and achieve greater organization, logical flow, and consistency as measured by GPT-4.1.

Correcting errors in long-video understanding is disproportionately costly: existing multimodal pipelines produce opaque, end-to-end outputs that expose no intermediate state for inspection, forcing annotators to revisit raw video and reconstruct temporal logic from scratch. The core bottleneck is not generation quality alone, but the absence of a supervisory interface through which human effort can be proportional to the scope of each error. We present IMPACT-CYCLE, a supervisory multi-agent system that reformulates long-video understanding as iterative claim-level maintenance of a shared semantic memory -- a structured, versioned state encoding typed claims, a claim dependency graph, and a provenance log. Role-specialized agents operating under explicit authority contracts decompose verification into local object-relation correctness, cross-temporal consistency, and global semantic coherence, with corrections confined to structurally dependent claims. When automated evidence is insufficient, the system escalates to human arbitration as the supervisory authority with final override rights; dependency-closure re-verification then ensures correction cost remains proportional to error scope. Experiments on VidOR show substantially improved downstream reasoning (VQA: 0.71 to 0.79) and a 4.8x reduction in human arbitration cost, with workload significantly lower than manual annotation. Code will be released at https://github.com/MKong17/IMPACT_CYCLE.