Distributed optimization is fundamental to large-scale machine learning and control applications. Among existing methods, the alternating direction method of multipliers (ADMM) has gained popularity due to its strong convergence guarantees and suitability for decentralized computation. However, ADMM can suffer from slow convergence and high sensitivity to hyperparameter choices. In this work, we show that distributed ADMM iterations can be naturally expressed within the message-passing framework of graph neural networks (GNNs). Building on this connection, we propose learning adaptive step sizes and communication weights through a GNN that predicts these yperparameters based on the current iterates. By unrolling ADMM for a fixed number of iterations, we train the network end-to-end to minimize the solution distance after these iterations for a given problem class, while preserving the algorithm's convergence properties. Numerical experiments demonstrate that our learned variant consistently improves convergence speed and solution quality compared to standard ADMM, both within the trained computational budget and beyond. The code is available at https://github.com/paulhausner/learning-distributed-admm.

Memory decay makes it harder for the human brain to recognize visual objects and retain details. Consequently, recorded brain signals become weaker, uncertain, and contain poor visual context over time. This paper presents one of the first vision-learning approaches to address this problem. First, we statistically and experimentally demonstrate the existence of inconsistency in brain signals and its impact on the Vision-Brain Understanding (VBU) model. Our findings show that brain signal representations shift over recording sessions, leading to compounding bias, which poses challenges for model learning and degrades performance. Then, we propose a new Bias-Mitigation Continual Learning (BRAIN) approach to address these limitations. In this approach, the model is trained in a continual learning setup and mitigates the growing bias from each learning step. A new loss function named De-bias Contrastive Learning is also introduced to address the bias problem. In addition, to prevent catastrophic forgetting, where the model loses knowledge from previous sessions, the new Angular-based Forgetting Mitigation approach is introduced to preserve learned knowledge in the model. Finally, the empirical experiments demonstrate that our approach achieves State-of-the-Art (SOTA) performance across various benchmarks, surpassing prior and non-continual learning methods.

A community of researchers appears to think that a machine can be surprised and have introduced various surprise measures, principally the Shannon Surprise and the Bayesian Surprise. The questions of what constitutes a surprise and how to react to one still elicit debates. In this work, we introduce Mutual Information Surprise (MIS), a new framework that redefines surprise not as anomaly measure, but as a signal of epistemic growth. Furthermore, we develop a statistical test sequence that could trigger a surprise reaction and propose a MIS-based reaction policy that dynamically governs system behavior through sampling adjustment and process forking. Empirical evaluations -- on both synthetic domains and a dynamic pollution map estimation task -- show that a system governed by the MIS-based reaction policy significantly outperforms those under classical surprise-based approaches in stability, responsiveness, and predictive accuracy. The important implication of our new proposal is that MIS quantifies the impact of new observations on mutual information, shifts surprise from reactive to reflective, enables reflection on learning progression, and thus offers a path toward self-aware and adaptive autonomous systems. We expect the new surprise measure to play a critical role in further advancing autonomous systems on their ability to learn and adapt in a complex and dynamic environment.

Infectious disease forecasting in novel outbreaks or low-resource settings is hampered by the need for large disease and covariate data sets, bespoke training, and expert tuning, all of which can hinder rapid generation of forecasts for new settings. To help address these challenges, we developed Mantis, a foundation model trained entirely on mechanistic simulations, which enables out-of-the-box forecasting across diseases, regions, and outcomes, even in settings with limited historical data. We evaluated Mantis against 78 forecasting models across sixteen diseases with diverse modes of transmission, assessing both point forecast accuracy (mean absolute error) and probabilistic performance (weighted interval score and coverage). Despite using no real-world data during training, Mantis achieved lower mean absolute error than all models in the CDC's COVID-19 Forecast Hub when backtested on early pandemic forecasts which it had not previously seen. Across all other diseases tested, Mantis consistently ranked in the top two models across evaluation metrics. Mantis further generalized to diseases with transmission mechanisms not represented in its training data, demonstrating that it can capture fundamental contagion dynamics rather than memorizing disease-specific patterns. These capabilities illustrate that purely simulation-based foundation models such as Mantis can provide a practical foundation for disease forecasting: general-purpose, accurate, and deployable where traditional models struggle.

Likelihood-based deep generative models have been widely investigated for Image Anomaly Detection (IAD), particularly Normalizing Flows, yet their strict architectural invertibility needs often constrain scalability, particularly in large-scale data regimes. Although time-parameterized Flow Matching (FM) serves as a scalable alternative, it remains computationally challenging in IAD due to the prohibitive costs of Jacobian-trace estimation. This paper proposes time-reversed Flow Matching (rFM), which shifts the objective from exact likelihood computation to evaluating target-domain regularity through density proxy estimation. We uncover two fundamental theoretical bottlenecks in this paradigm: first, the reversed vector field exhibits a non-Lipschitz singularity at the initial temporal boundary, precipitating explosive estimation errors. Second, the concentration of measure in high-dimensional Gaussian manifolds induces structured irregularities, giving rise to a Centripetal Potential Field (CPF) that steers trajectories away from Optimal Transport (OT) paths. We identify these observations as the inherent dualities between FM and rFM. To address these issues, we introduce local Worst Transport Flow matching (WT-Flow), which amplifies the observed CPF of rFM to mitigate the initial singularity while circumventing the need for exact distribution transformations via density proxy. Experiments on five datasets demonstrate that WT-Flow achieves state-of-the-art performance among single-scale flow-based methods, and competitive performance against leading multi-scale approaches. Furthermore, the proposed framework enables superior one-step inference, achieving a per-image flow latency of only 6.7 ms. Our code is available on https://github.com/lil-wayne-0319/fmad.

Machine Unlearning (MU) aims to selectively erase the influence of specific data points from pretrained models. However, most existing MU methods rely on the retain set to preserve model utility, which is often impractical due to privacy restrictions and storage constraints. While several retain-data-free methods attempt to bypass this using geometric feature shifts or auxiliary statistics, they typically treat forgetting samples uniformly, overlooking their heterogeneous contributions. To address this, we propose \ul{I}nfluence-guided \ul{M}achine \ul{U}nlearning (IMU), a principled method that conducts MU using only the forget set. Departing from uniform Gradient Ascent (GA) or implicit weighting mechanisms, IMU leverages influence functions as an explicit priority signal to allocate unlearning strength. To circumvent the prohibitive cost of full-model Hessian inversion, we introduce a theoretically grounded classifier-level influence approximation. This efficient design allows IMU to dynamically reweight unlearning updates, aggressively targeting samples that most strongly support the forgetting objective while minimizing unnecessary perturbation to retained knowledge. Extensive experiments across vision and language tasks show that IMU achieves highly competitive results. Compared to standard uniform GA, IMU maintains identical unlearning depth while enhancing model utility by an average of 30%, effectively overcoming the inherent utility-forgetting trade-off.

Obtaining human-readable symbolic formulas via genetic programming-based symbolic distillation of a deep neural network trained on the target dataset presents a promising yet underexplored path towards explainable artificial intelligence (XAI); however, the standard pipeline frequently yields symbolic models with poor predictive accuracy. We identify a fundamental misalignment in functional complexity as the primary barrier to achieving better accuracy: standard Artificial Neural Networks (ANNs) often learn accurate but highly irregular functions, while Symbolic Regression typically prioritizes parsimony, often resulting in a much simpler class of models that are unable to sufficiently distill or learn from the ANN teacher. To bridge this gap, we propose a framework that actively regularizes the teacher's functional smoothness using Jacobian and Lipschitz penalties, aiming to distill better student models than the standard pipeline. We characterize the trade-off between predictive accuracy and functional complexity through a robust study involving 20 datasets and 50 independent trials. Our results demonstrate that students distilled from smoothness-regularized teachers achieve statistically significant improvements in R^2 scores, compared to the standard pipeline. We also perform ablation studies on the student model algorithm. Our findings suggest that smoothness alignment between teacher and student models is a critical factor for symbolic distillation.

Real-time trajectory planning for unmanned aerial vehicles (UAVs) in dynamic environments remains a key challenge due to high computational demands and the need for fast, adaptive responses. Traditional Particle Swarm Optimization (PSO) methods, while effective for offline planning, often struggle with premature convergence and latency in real-time scenarios. To overcome these limitations, we propose PE-PSO, an enhanced PSO-based online trajectory planner. The method introduces a persistent exploration mechanism to preserve swarm diversity and an entropy-based parameter adjustment strategy to dynamically adapt optimization behavior. UAV trajectories are modeled using B-spline curves, which ensure path smoothness while reducing optimization complexity. To extend this capability to UAV swarms, we develop a multi-agent framework that combines genetic algorithm (GA)-based task allocation with distributed PE-PSO, supporting scalable and coordinated trajectory generation. The distributed architecture allows for parallel computation and decentralized control, enabling effective cooperation among agents while maintaining real-time performance. Comprehensive simulations demonstrate that the proposed framework outperforms conventional PSO and other swarm-based planners across several metrics, including trajectory quality, energy efficiency, obstacle avoidance, and computation time. These results confirm the effectiveness and applicability of PE-PSO in real-time multi-UAV operations under complex environmental conditions.

Ground penetrating radar (GPR) has become a rapid and non-destructive solution for road subsurface distress (RSD) detection. However, recognizing RSD from GPR images is labor-intensive and heavily relies on the expertise of inspectors. Deep learning-based automatic RSD recognition, though ameliorating the burden of data processing, suffers from insufficient capability to recognize defects. In this study, a novel cross-verification strategy was proposed to fully exploit the complementary abilities of region proposal networks in object recognition from different views of GPR images. Following this strategy, three YOLO-based models were used to detect the RSD (voids and loose structures) and manholes. Each model was trained with a specific view of 3D GPR dataset, which contains rigorously validated 2134 samples of diverse types obtained through field scanning. The cross-verification strategy achieves outstanding accuracy with a recall of over 98.6% in the tests using real field-scanning data. Field tests also show that deep learning-based automatic RSD recognition can reduce the human labor of inspection by around 90%.

Split learning (SL) is a distributed learning paradigm that can enable computation-intensive artificial intelligence (AI) applications by partitioning AI models between mobile devices and edge servers. %fully utilizing distributed computing resources for computation-intensive mobile intelligence applications. However, the model partitioning problem in SL becomes challenging due to the diverse and complex architectures of AI models. In this paper, we formulate an optimal model partitioning problem to minimize training delay in SL. To solve the problem, we represent an arbitrary AI model as a directed acyclic graph (DAG), where the model's layers and inter-layer connections are mapped to vertices and edges, and training delays are captured as edge weights. Then, we propose a general model partitioning algorithm by transforming the problem into a minimum \textit{s-t} cut problem on the DAG. Theoretical analysis shows that the two problems are equivalent, such that the optimal model partition can be obtained via a maximum-flow method. Furthermore, taking AI models with block structures into consideration, we design a low-complexity block-wise model partitioning algorithm to determine the optimal model partition. Specifically, the algorithm simplifies the DAG by abstracting each block (i.e., a repeating component comprising multiple layers in an AI model) into a single vertex. Extensive experimental results on a hardware testbed equipped with NVIDIA Jetson devices demonstrate that the proposed solution can reduce algorithm running time by up to 13.0$\times$ and training delay by up to 38.95\%, compared to state-of-the-art baselines.

Interactive segmentation (IS) allows users to iteratively refine object boundaries with minimal cues, such as positive and negative clicks. While the Segment Anything Model (SAM) has garnered attention in the IS community for its promptable segmentation capabilities, it often struggles in specialized domains or when handling complex scenarios (e.g., camouflaged or multi-part objects). To overcome these challenges, we propose DC-TTA, a novel test-time adaptation (TTA) framework that adapts SAM on a per-sample basis by leveraging user interactions as supervision. Instead of forcing a single model to incorporate all user clicks at once, DC-TTA partitions the clicks into more coherent subsets, each processed independently via TTA with a separated model. This Divide-and-Conquer strategy reduces conflicts among diverse cues and enables more localized updates. Finally, we merge the adapted models to form a unified predictor that integrates the specialized knowledge from each subset. Experimental results across various benchmarks demonstrate that DC-TTA significantly outperforms SAM's zero-shot results and conventional TTA methods, effectively handling complex tasks such as camouflaged object segmentation with fewer interactions and improved accuracy.

Large Language Models (LLMs) have undergone rapid progress, largely attributed to reinforcement learning on complex reasoning tasks. In contrast, while spatial intelligence is fundamental for Vision-Language Models (VLMs) in real-world interaction, the systematic study of their complex spatial reasoning remains underexplored. To bridge this gap, we introduce SIRI-Bench, a benchmark designed to evaluate VLMs' structural spatial intelligence through spatial-grounded reasoning tasks. SIRI-Bench comprises 9,000 video-question-answer triplets, where each problem is embedded in a realistic 3D scene. The benchmark is carefully designed so that solving each problem requires both spatial comprehension and structural reasoning. To facilitate large-scale data synthesis, we develop an Automatic Scene Creation Engine that employs collaborative LLM agents to translate abstract mathematical problems into faithful 3D scenes. Experimental results reveal that state-of-the-art VLMs struggle significantly on SIRI-Bench, underscoring the challenge of structural spatial reasoning. We hope that our study will bring researchers' attention to spatially grounded reasoning and advance VLMs in visual problem-solving.

Large language models (LLMs) are increasingly deployed in decision-support systems for high-stakes domains such as hiring and university admissions, where choices often involve selecting among competing alternatives. While prior work has noted position biases in LLM-driven comparisons, these biases have not been systematically analyzed or linked to underlying preference structures. We present the first comprehensive study of position biases across multiple LLMs and two distinct domains: resume comparisons, representing a realistic high-stakes context, and color selection, which isolates position effects by removing confounding factors. We find strong and consistent order effects, including a quality-dependent shift: when all options are high quality, models favor the first option, but when quality is lower, they favor later options. We also identify a previously undocumented bias: a name bias, where certain names are favored despite controlling for demographic signals. To separate superficial tie-breaking from genuine distortions of judgment, we extend the rational choice framework to classify pairwise preferences as robust, fragile, or indifferent. Using this framework, we show that order effects can lead models to select strictly inferior options. These results indicate that LLMs exhibit distinct failure modes not documented in human decision-making. We also propose targeted mitigation strategies, including a novel use of the temperature parameter, to recover underlying preferences when order effects distort model behavior.

The ability of AI to sense and identify various substances based on their smell alone can have profound impacts on allergen detection (e.g. smelling gluten or peanuts in a cake), monitoring the manufacturing process, and sensing hormones that indicate emotional states, stress levels, and diseases. Despite these broad impacts, there are few standardized datasets, and therefore little progress, for training and evaluating AI systems' ability to `smell' in the real-world. In this paper, we use small gas and chemical sensors to create SmellNet, a comparatively large dataset for sensor-based machine olfaction that digitizes a diverse range of smells in the natural world. SmellNet contains about 828,000 time-series data points across 50 substances, spanning nuts, spices, herbs, fruits, and vegetables, and 43 mixtures among them with fixed ingredient volumetric ratios, with 68 hours of data collected. Using SmellNet, we developed ScentFormer, a Transformer-based architecture combining temporal differencing and sliding-window augmentation for smell data. For the SmellNet-Base classification tasks, ScentFormer achieves 63.3% Top-1 accuracy with GC-MS supervision, and for the SmellNet-Mixture distribution prediction tasks, ScentFormer achieves 50.2% Top-1@0.1 on the test-seen split. ScentFormer's ability to generalize across conditions and capture transient chemical dynamics demonstrates the promise of temporal modeling in sensor-based olfactory AI. SmellNet and ScentFormer lay the groundwork for sensor-based olfactory applications across healthcare, food and beverage, environmental monitoring, manufacturing, and entertainment.

Current text-to-image diffusion generation typically employs complete-text conditioning. Due to the intricate syntax, diffusion transformers (DiTs) inherently suffer from a comprehension defect of complete-text captions. One-fly complete-text input either overlooks critical semantic details or causes semantic confusion by simultaneously modeling diverse semantic primitive types. To mitigate this defect of DiTs, we propose a novel split-text conditioning framework named DiT-ST. This framework converts a complete-text caption into a split-text caption, a collection of simplified sentences, to explicitly express various semantic primitives and their interconnections. The split-text caption is then injected into different denoising stages of DiT-ST in a hierarchical and incremental manner. Specifically, DiT-ST leverages Large Language Models to parse captions, extracting diverse primitives and hierarchically sorting out and constructing these primitives into a split-text input. Moreover, we partition the diffusion denoising process according to its differential sensitivities to diverse semantic primitive types and determine the appropriate timesteps to incrementally inject tokens of diverse semantic primitive types into input tokens via cross-attention. In this way, DiT-ST enhances the representation learning of specific semantic primitive types across different stages. Extensive experiments validate the effectiveness of our proposed DiT-ST in mitigating the complete-text comprehension defect.

Large Language Models (LLMs) equipped with modern Retrieval-Augmented Generation (RAG) systems often employ multi-turn interaction pipelines to interface with search engines for complex reasoning tasks. However, such multi-turn interactions inevitably produce long intermediate contexts, as context length grows exponentially with exploration depth. This leads to a well-known limitation of LLMs: their difficulty in effectively leveraging information from long contexts. This problem is further amplified in RAG systems that depend on in-context learning, where few-shot demonstrations must also be included in the prompt, compounding the context-length bottleneck. To address these challenges, we propose Mujica-MyGo, a unified framework for efficient multi-turn reasoning in RAG. Inspired by the divide-and-conquer principle, we introduce Mujica (Multi-hop Joint Intelligence for Complex Question Answering), a multi-agent RAG workflow that decomposes multi-turn interactions into cooperative sub-interactions, thereby mitigating long-context issues. To eliminate the dependency on in-context learning, we further develop MyGO (Minimalist Policy Gradient Optimization), a lightweight and efficient reinforcement learning algorithm that enables effective post-training of LLMs within complex RAG pipelines. We provide theoretical guarantees for MyGO's convergence to the optimal policy. Empirical evaluations across diverse question-answering benchmarks, covering both text corpora and knowledge graphs, show that Mujica-MyGO achieves superior performance.

Large language models have achieved remarkable success in various tasks. However, it is challenging for them to learn new tasks incrementally due to catastrophic forgetting. Existing approaches rely on experience replay, optimization constraints, or task differentiation, which encounter strict limitations in real-world scenarios. To address these issues, we propose Joint Flashback Adaptation. We first introduce flashbacks -- a limited number of prompts from old tasks -- when adapting to new tasks and constrain the deviations of the model outputs compared to the original one. We then interpolate latent tasks between flashbacks and new tasks to enable jointly learning relevant latent tasks, new tasks, and flashbacks, alleviating data sparsity in flashbacks and facilitating knowledge sharing for smooth adaptation. Our method requires only a limited number of flashbacks without access to the replay data and is task-agnostic. We conduct extensive experiments on state-of-the-art large language models across 1000+ instruction-following tasks, arithmetic reasoning tasks, and general reasoning tasks. The results demonstrate the superior performance of our method in improving generalization on new tasks and reducing forgetting in old tasks.

Reinforcement learning (RL) has emerged as an effective approach for enhancing the reasoning capabilities of large language models (LLMs), especially in scenarios where supervised fine-tuning (SFT) falls short due to limited chain-of-thought (CoT) data. Among RL-based post-training methods, group relative advantage estimation, as exemplified by Group Relative Policy Optimization (GRPO), has attracted considerable attention for eliminating the dependency on the value model, thereby simplifying training compared to traditional approaches like Proximal Policy Optimization (PPO). However, we observe that exsiting group relative advantage estimation method still suffers from training inefficiencies, particularly when the estimated advantage approaches zero. To address this limitation, we propose Advantage-Augmented Policy Optimization (AAPO), a novel RL algorithm that optimizes the cross-entropy (CE) loss using advantages enhanced through a margin-based estimation scheme. This approach effectively mitigates the inefficiencies associated with group relative advantage estimation. Experimental results on multiple mathematical reasoning benchmarks demonstrate the superior performance of AAPO. Code is available at https://github.com/JianxXiong/AAPO.

Evolution and learning have historically been interrelated topics, and their interplay is attracting increased interest lately. The emerging new factor in this trend is morphological evolution, the evolution of physical forms within embodied AI systems such as robots. In this study, we investigate a system of hexacopter-type drones with evolvable morphologies and learnable controllers and make contributions to two fields. For aerial robotics, we demonstrate that the combination of evolution and learning can deliver non-conventional drones that significantly outperform the traditional hexacopter on several tasks that are more complex than previously considered in the literature. For the field of Evolutionary Computing, we introduce novel metrics and perform new analyses into the interaction of morphological evolution and learning, uncovering hitherto unidentified effects. Our analysis tools are domain-agnostic, making a methodological contribution towards building solid foundations for embodied AI systems that integrate evolution and learning.

We study the geometry of Receiver Operating Characteristic (ROC) and Precision-Recall (PR) curves in binary classification problems. The key finding is that many of the most commonly used binary classification metrics are merely functions of the composition function $G := F_p \circ F_n^{-1}$, where $F_p(\cdot)$ and $F_n(\cdot)$ are the class-conditional cumulative distribution functions of the classifier scores in the positive and negative classes, respectively. This geometric perspective facilitates the selection of operating points, understanding the effect of decision thresholds, and comparison between classifiers. It also helps explain how the shapes and geometry of ROC/PR curves reflect classifier behavior, providing objective tools for building classifiers optimized for specific applications with context-specific constraints. We further explore the conditions for classifier dominance, present analytical and numerical examples demonstrating the effects of class separability and variance on ROC and PR geometries, and derive a link between the positive-to-negative class leakage function $G(\cdot)$ and the Kullback-Leibler divergence. The framework highlights practical considerations, such as model calibration, cost-sensitive optimization, and operating point selection under real-world capacity constraints, enabling more informed approaches to classifier deployment and decision-making.

Weakly supervised object localization (WSOL) methods allow training models to classify images and localize ROIs. WSOL only requires low-cost image-class annotations yet provides a visually interpretable classifier. Standard WSOL methods rely on class activation mapping (CAM) methods to produce spatial localization maps according to a single- or two-step strategy. While both strategies have made significant progress, they still face several limitations with histology images. Single-step methods can easily result in under- or over-activation due to the limited visual ROI saliency in histology images and scarce localization cues. They also face the well-known issue of asynchronous convergence between classification and localization tasks. The two-step approach is sub-optimal because it is constrained to a frozen classifier, limiting the capacity for localization. Moreover, these methods also struggle when applied to out-of-distribution (OOD) datasets. In this paper, a multi-task approach for WSOL is introduced for simultaneous training of both tasks to address the asynchronous convergence problem. In particular, localization is performed in the pixel-feature space of an image encoder that is shared with classification. This allows learning discriminant features and accurate delineation of foreground/background regions to support ROI localization and image classification. We propose PixelCAM, a cost-effective foreground/background pixel-wise classifier in the pixel-feature space that allows for spatial object localization. Using partial-cross entropy, PixelCAM is trained using pixel pseudo-labels collected from a pretrained WSOL model. Both image and pixel-wise classifiers are trained simultaneously using standard gradient descent. In addition, our pixel classifier can easily be integrated into CNN- and transformer-based architectures without any modifications.

Brains construct not only "first-order" representations of the environment but also "higher-order" representations about those representations -- including higher-order uncertainty estimates that guide learning and adaptive behavior. Higher-order expectations about representational uncertainty -- i.e., learned through experience -- may play a key role in guiding behavior and learning, but their characterization remains empirically and theoretically challenging. Here, we introduce the Noise Estimation through Reinforcement-based Diffusion (NERD) model, a novel computational framework that trains denoising diffusion models via reinforcement learning to infer distributions of noise in functional MRI data from a decoded neurofeedback task, where healthy human participants learn to achieve target neural states. We hypothesize that participants accomplish this task by learning about and then minimizing their own representational uncertainty. We test this hypothesis with NERD, which mirrors brain-like unsupervised learning. Our results show that NERD outperforms backpropagation-trained control models in capturing human performance with explanatory power enhanced by clustering learned noise distributions. Importantly, our results also reveal individual differences in expected-uncertainty representations that predict task success, demonstrating NERD's utility as a powerful tool for probing higher-order neural representations.

We study the efficacy of fine-tuning Large Language Models (LLMs) for the specific task of report (government archives, news, intelligence reports) summarization. While this topic is being very actively researched - our specific application set-up faces two challenges: (i) ground-truth summaries maybe unavailable (e.g., for government archives), and (ii) availability of limited compute power - the sensitive nature of the application requires that computation is performed on-premise and for most of our experiments we use one or two A100 GPU cards. Under this set-up we conduct experiments to answer the following questions. First, given that fine-tuning the LLMs can be resource intensive, is it feasible to fine-tune them for improved report summarization capabilities on-premise? Second, what are the metrics we could leverage to assess the quality of these summaries? We conduct experiments on two different fine-tuning approaches in parallel and our findings reveal interesting trends regarding the utility of fine-tuning LLMs. Specifically, we find that in many cases, fine-tuning helps improve summary quality and in other cases it helps by reducing the number of invalid or garbage summaries.

Extreme hazard events such as wildfires and hurricanes increasingly threaten power systems, causing widespread outages and disrupting critical services. Recently, predict-then-optimize approaches have gained traction in grid operations, where system functionality forecasts are first generated and then used as inputs for downstream decision-making. However, this two-stage method often results in a misalignment between prediction and optimization objectives, leading to suboptimal resource allocation. To address this, we propose predict-all-then-optimize-globally (PATOG), a framework that integrates outage prediction with globally optimized interventions. At its core, our global-decision-focused (GDF) neural ODE model captures outage dynamics while optimizing resilience strategies in a decision-aware manner. Unlike conventional methods, our approach ensures spatially and temporally coherent decision-making, improving both predictive accuracy and operational efficiency. Experiments on synthetic and real-world datasets demonstrate significant improvements in outage prediction consistency and grid resilience.

Electronic Health Records (EHRs) offer considerable potential for clinical prediction, but their complexity and heterogeneity challenge traditional machine learning. Domain-specific EHR foundation models trained on unlabeled EHR data have shown improved predictive accuracy and generalization. However, their development is constrained by limited data access and site-specific vocabularies. We convert EHR data into plain text by replacing medical codes with natural-language descriptions, enabling general-purpose Large Language Models (LLMs) to produce high-dimensional embeddings for downstream prediction tasks without access to private medical training data. LLM-based embeddings perform on par with a specialized EHR foundation model, CLMBR-T-Base, across 15 clinical tasks from the EHRSHOT benchmark. In an external validation using the UK Biobank, an LLM-based model shows statistically significant improvements for some tasks, which we attribute to higher vocabulary coverage and slightly better generalization. Overall, we reveal a trade-off between the computational efficiency of specialized EHR models and the portability and data independence of LLM-based embeddings.

Solving complex reasoning tasks may involve visual understanding, domain knowledge retrieval, numerical calculation, and multi-step reasoning. Existing methods augment large language models (LLMs) with external tools but are restricted to specialized domains, limited tool types, or require additional training data. In this paper, we introduce OctoTools, a training-free, user-friendly, and easily extensible multi-agent framework designed to tackle complex reasoning across diverse domains. OctoTools introduces standardized tool cards to encapsulate tool functionality, a planner for both high-level and low-level planning, and an executor to carry out tool usage. We validate OctoTools' generality across 16 diverse tasks (including MathVista, MMLU-Pro, MedQA, and GAIA-Text), achieving substantial average accuracy gains of 9.3% over GPT-4o. Furthermore, OctoTools also outperforms AutoGen, GPT-Functions, and LangChain by up to 10.6% when given the same set of tools. Through comprehensive analysi, ablations, and robustness tests with compact backbones and noisy tool environments, OctoTools demonstrates advantages in task planning, effective tool usage, and multi-step problem solving. Code, demos, and visualization are publicly available at https://octotools.github.io/.

Hierarchical data is common in many domains like life sciences and e-commerce, and its embeddings often play a critical role. While hyperbolic embeddings offer a theoretically grounded approach to representing hierarchies in low-dimensional spaces, current methods often rely on specific geometric constructs as embedding candidates. This reliance limits their generalizability and makes it difficult to integrate with techniques that model semantic relationships beyond pure hierarchies, such as ontology embeddings. In this paper, we present RegD, a flexible Euclidean framework that supports the use of arbitrary geometric regions -- such as boxes and balls -- as embedding representations. Although RegD operates entirely in Euclidean space, we formally prove that it achieves hyperbolic-like expressiveness by incorporating a depth-based dissimilarity between regions, enabling it to emulate key properties of hyperbolic geometry, including exponential growth. Our empirical evaluation on diverse real-world datasets shows consistent performance gains over state-of-the-art methods and demonstrates RegD's potential for broader applications such as the ontology embedding task that goes beyond hierarchy.

Large language models (LLMs) have shown impressive multilingual capabilities through pretraining on diverse corpora. Although these models show strong reasoning abilities, their performance varies significantly between languages due to the imbalanced distribution of training data. Existing approaches using sample-level translation for extensive multilingual pretraining and cross-lingual tuning face scalability challenges and often fail to capture nuanced reasoning processes across languages. In this paper, we introduce AdaMCOT (Adaptive Multilingual Chain-of-Thought), a framework that enhances multilingual factual reasoning by dynamically routing thought processes in intermediary "thinking languages" before generating target-language responses. AdaMCOT leverages a language-agnostic core and incorporates an adaptive, reward-based mechanism for selecting optimal reasoning pathways without requiring additional pretraining. Our comprehensive evaluation across multiple benchmarks demonstrates substantial improvements in both factual reasoning quality and cross-lingual consistency, with particularly strong performance gains in low-resource language settings. An in-depth analysis of the model's hidden states and semantic space further elucidates the underlying mechanism of our method. The results suggest that adaptive reasoning paths can effectively bridge the performance gap between high and low-resource languages while maintaining cultural and linguistic nuances.

We propose a graph-based clustering method based on Cluster Catch Digraphs (CCDs) that extends their applicability to moderate-dimensional data settings. Existing CCD variants, such as RK-CCDs, rely on spatial randomness tests based on Ripley's K function, which exhibit performance degradation as dimensionality increases. To address this limitation, we introduce a nearest-neighbor-distance (NND) based Monte Carlo spatial randomness test (MC-SRT) for determining covering radii, resulting in the proposed Uniformity- and Neighbor-based CCDs (UN-CCDs). The proposed method is designed for datasets of moderate size and dimension, particularly in settings with complex cluster geometry and uniformly distributed background noise. Through Monte Carlo simulations and experiments on benchmark datasets, we show that UN-CCDs provide stable and competitive performance relative to several established clustering methods within the evaluated regimes, while remaining largely parameter-free. We also discuss computational trade-offs and identify the practical regimes in which the method is most effective. -- Keywords: Graph-based clustering; Cluster catch digraphs; Moderate-dimensional data; the nearest neighbor distance; Spatial randomness test.

Most Graph Neural Networks (GNNs) operate at the first-order scale, even though multi-scale representations are known to be crucial in domains such as image classification. In this work, we investigate whether GNNs can similarly benefit from multi-scale learning, rather than being limited to a fixed depth of $k$-hop aggregation. We begin by formalizing scale invariance in graph learning, providing theoretical guarantees and empirical evidence for its effectiveness. Building on this principle, we introduce ScaleNet, a scale-aware message-passing architecture that combines directed multi-scale feature aggregation with an adaptive self-loop mechanism. ScaleNet achieves state-of-the-art performance on six benchmark datasets, covering both homophilic and heterophilic graphs. To handle scalability, we further propose LargeScaleNet, which extends multi-scale learning to large graphs and sets new state-of-the-art results on three large-scale benchmarks. We also show that FaberNet's strength largely arises from multi-scale feature integration. Together with these state-of-the-art results, our findings suggest that scale invariance may serve as a valuable principle for improving the performance of single-order GNNs. The code for all experiments is available at \href{https://github.com/Qin87/ScaleNet/tree/iclr_scale_aware/}{this link}.