NeurIPS 2025 Papers

We study the distributed multi-agent multi-armed bandit problem with heterogeneous rewards over random communication graphs. Uniquely, at each time step $t$ agents communicate over a time-varying random graph $\mathcal{G}\_t$ generated by applying the Erdős–Rényi model to a fixed connected base graph $\mathcal{G}$ (for classical Erdos-Rényi graphs, $\mathcal{G}$ is a complete graph), where each potential edge in $\mathcal{G}$ is randomly and independently present with the link probability $p$. Notably, the resulting random graph is not necessarily connected at each time step. Each agent's arm rewards follow time-invariant distributions, and the reward distribution for the same arm may differ across agents. The goal is to minimize the cumulative expected regret relative to the global mean reward of each arm, defined as the average of that arm’s mean rewards across all agents. To this end, we propose a fully distributed algorithm that integrates the arm elimination strategy with the random gossip algorithm. We theoretically show that the regret upper bound is of order $\log T$ and is highly interpretable, where $T$ is the time horizon. It includes the optimal centralized regret $\mathcal O\left(\sum_{k: \Delta_k>0} \frac{\log T}{\Delta_k}\right)$ and an additional term $\mathcal O\left(\frac{N^2 \log T}{p \lambda_{N-1}(\operatorname{Lap}(\mathcal{G}))} + \frac{KN^2 \log T}{p}\right)$ where $N$ and $K$ denote the total number of agents and arms, respectively. This term reflects the impact of $\mathcal G$'s algebraic connectivity $\lambda_{N-1}(\operatorname{Lap}(\mathcal{G}))$ and the link probability $p$, and thus highlights a fundamental trade-off between communication efficiency and regret. As a by-product, we show a nearly optimal regret lower bound. Finally, our numerical experiments not only show the superiority of our algorithm over existing benchmarks, but also validate the theoretical regret scaling with problem complexity.

We address the challenge of adopting language models (LMs) for embodied tasks in dynamic environments, where online access to large-scale inference engines or symbolic planners is constrained due to latency, connectivity, and resource limitations. To this end, we present NeSyPr, a novel embodied reasoning framework that compiles knowledge via neurosymbolic proceduralization, thereby equipping LM-based agents with structured, adaptive, and timely reasoning capabilities. In NeSyPr, task-specific plans are first explicitly generated by a symbolic tool leveraging its declarative knowledge. These plans are then transformed into composable procedural representations that encode the plans' implicit production rules, enabling the resulting composed procedures to be seamlessly integrated into the LM's inference process. This neurosymbolic proceduralization abstracts and generalizes multi-step symbolic structured path-finding and reasoning into single-step LM inference, akin to human knowledge compilation. It supports efficient test-time inference without relying on external symbolic guidance, making it well suited for deployment in latency-sensitive and resource-constrained physical systems. We evaluate NeSyPr on the embodied benchmarks PDDLGym, VirtualHome, and ALFWorld, demonstrating its efficient reasoning capabilities over large-scale reasoning models and a symbolic planner, while using more compact LMs.

Machine learning is now ubiquitous in societal decision-making, for example in evaluating job candidates or loan applications, and it is increasingly important to take into account how classified agents will react to the learning algorithms. The majority of recent literature on strategic classification has focused on reducing and countering deceptive behaviors by the classified agents, but recent work of Attias et al. identifies surprising properties of learnability when the agents genuinely improve in order to attain the desirable classification, such as smaller generalization error than standard PAC-learning. In this paper we characterize so-called learnability with improvements across multiple new axes. We introduce an asymmetric variant of minimally consistent concept classes and use it to provide an exact characterization of proper learning with improvements in the realizable setting. While prior work studies learnability only under general, arbitrary agent improvement regions, we give positive results for more natural Euclidean ball improvement sets. In particular, we characterize improper learning under a generative assumption on the data distribution. We further show how to learn in more challenging settings, achieving lower generalization error under well-studied bounded noise models and obtaining mistake bounds in realizable and agnostic online learning. We resolve open questions posed by Attias et al. for both proper and improper learning.

Swimming in nature achieves remarkable performance through diverse morphological adaptations and intricate solid-fluid interaction, yet exploring this capability in artificial soft swimmers remains challenging due to the high-dimensional control complexity and the computational cost of resolving hydrodynamic details. Traditional approaches often rely on morphology-dependent heuristics and simplified fluid models, which constrain exploration and preclude advanced strategies like vortex exploitation. To address this, we propose an automated framework that combines a unified, reduced-mode control space with a high-fidelity GPU-accelerated simulator. Our control space naturally captures deformation patterns for diverse morphologies, minimizing manual design, while our simulator efficiently resolves the crucial fluid-structure interactions required for learning. We evaluate our method on a wide range of morphologies, from bio-inspired to unconventional. From this general framework, high-performance swimming patterns emerge that qualitatively reproduce canonical gaits observed in nature without requiring domain-specific priors, where state-of-the-art baselines often fail, particularly on complex topologies like a torus. Our work lays a foundation for future opportunities in automated co-design of soft robots in complex hydrodynamic environments. The code is available at https://github.com/changyu-hu/FreeFlow.

Clinical decision making (CDM) is a complex, dynamic process crucial to healthcare delivery, yet it remains a significant challenge for artificial intelligence systems. While Large Language Model (LLM)-based agents have been tested on general medical knowledge using licensing exams and knowledge question-answering tasks, their performance in the CDM in real-world scenarios is limited due to the lack of comprehensive benchmark that mirror actual medical practice. To address this gap, we present MedChain, a dataset of 12,163 clinical cases that covers five key stages of clinical workflow. MedChain distinguishes itself from existing benchmarks with three key features of real-world clinical practice: personalization, interactivity, and sequentiality. Further, to tackle real-world CDM challenges, we also propose MedChain-Agent, an AI system that integrates a feedback mechanism and a MedCase-RAG module to learn from previous cases and adapt its responses. MedChain-Agent demonstrates remarkable adaptability in gathering information dynamically and handling sequential clinical tasks, significantly outperforming existing approaches. The relevant dataset and code will be released upon acceptance of this paper.

Public opinion governance in social networks is critical for public health campaigns, political elections, and commercial marketing. In this paper, we addresse the problem of maximizing overall opinion in social networks by strategically modifying the internal opinions of key nodes. Traditional matrix inversion methods suffer from prohibitively high computational costs, prompting us to propose two efficient sampling-based algorithms. Furthermore, we develop a deterministic asynchronous algorithm that exactly identifies the optimal set of nodes through asynchronous update operations and progressive refinement, ensuring both efficiency and precision. Extensive experiments on real-world datasets demonstrate that our methods outperform baseline approaches. Notably, our asynchronous algorithm delivers exceptional efficiency and accuracy across all scenarios, even in networks with tens of millions of nodes.

Graphical User Interface (GUI) agents show promising capabilities for automating computer-use tasks and facilitating accessibility, but existing interactive benchmarks are mostly English-only, covering web-use or Windows, Linux, and Android environments, but not macOS. macOS is a major OS with distinctive GUI patterns and exclusive applications. To bridge the gaps, we present macOSWorld, the first comprehensive benchmark for evaluating GUI agents on macOS. macOSWorld features 202 multilingual interactive tasks across 30 applications (28 macOS-exclusive), with task instructions and OS interfaces offered in 5 languages (English, Chinese, Arabic, Japanese, and Russian). As GUI agents are shown to be vulnerable to deception attacks, macOSWorld also includes a dedicated safety benchmarking subset. Our evaluation on six GUI agents reveals a dramatic gap: proprietary computer-use agents lead at above 30\% success rate, while open-source lightweight research models lag at below 5\%, highlighting the need for macOS domain adaptation. Multilingual benchmarks also expose common weaknesses, especially in Arabic, with a 28.8\% average degradation compared to English. Results from safety benchmarking also highlight that deception attacks are more general and demand immediate attention. Project page: https://macos-world.github.io.

Driven by the applications in autonomous driving, robotics, and augmented reality, 3D object annotation is a critical task compared to 2D annotation, such as spatial complexity, occlusion, and viewpoint inconsistency. The existing methods relying on single models often struggle with these issues. In this paper, we introduce Tri-MARF, a novel framework that integrates tri-modal inputs (i.e., 2D multi-view images, text descriptions, and 3D point clouds) with multi-agent collaboration to enhance the 3D annotation process. Our Tri-MARF consists of three specialized agents: a vision-language model agent that generates multi-view descriptions, an information aggregation agent that selects optimal descriptions, and a gating agent that aligns text descriptions with 3D geometries for more refined captioning. Extensive experiments on the Objaverse-LVIS, Objaverse-XL, and ABO datasets demonstrate the superiority of our Tri-MARF, which achieves a CLIPScore of 88.7 (compared to 78.6–82.4 for other SOTA methods), retrieval accuracy of 45.2/43.8 (ViLT R@5), and an impressive throughput of 12,000 objects per hour on a single NVIDIA A100 GPU.

We propose a novel collaborative multi-agent optimization framework for adaptive training in multi-label image classification, fundamentally advancing beyond static decision rules and isolated automation. Our method deploys a set of distributed, task-specific agents, each responsible for dynamically orchestrating critical training components—including data augmentation, optimization methods, learning rate schedules, and loss functions—according to evolving visual-semantic relationships and training states. Each agent employs an advanced non-stationary multi-armed bandit algorithm, integrating both $\epsilon$-greedy and upper confidence bound strategies, to judiciously balance exploration with exploitation throughout the training lifecycle. A hierarchical composite reward mechanism synergizes overall classification accuracy, rare class recognition, and training stability, fostering both independent optimization and implicit collaborative behavior among agents. The framework further leverages refined techniques such as dual-rate exponential moving average smoothing and structured mixed-precision training to enhance robustness and computational efficiency. Extensive experiments across benchmarks including Pascal VOC, COCO, Yeast, and Mediamill demonstrate that our approach achieves superior mean average precision and rare-class F1 scores compared to state-of-the-art methods, while also exhibiting rapid convergence and remarkable cross-domain generalization. Our results indicate that collaborative multi-agent adaptive optimization offers a scalable and principled solution for self-optimizing deep learning in complex multi-label scenarios.

A key challenge in offline multi-agent reinforcement learning (MARL) is achieving effective many-agent multi-step coordination in complex environments. In this work, we propose Oryx, a novel algorithm for offline cooperative MARL to directly address this challenge. Oryx adapts the recently proposed retention-based architecture Sable and combines it with a sequential form of implicit constraint Q-learning (ICQ), to develop a novel offline autoregressive policy update scheme. This allows Oryx to solve complex coordination challenges while maintaining temporal coherence over long trajectories. We evaluate Oryx across a diverse set of benchmarks from prior works—SMAC, RWARE, and Multi-Agent MuJoCo—covering tasks of both discrete and continuous control, varying in scale and difficulty. Oryx achieves state-of-the-art performance on more than 80% of the 65 tested datasets, outperforming prior offline MARL methods and demonstrating robust generalisation across domains with many agents and long horizons. Finally, we introduce new datasets to push the limits of many-agent coordination in offline MARL, and demonstrate Oryx's superior ability to scale effectively in such settings.

Liquid cooling is critical for thermal management in high-density data centers with the rising AI workloads. However, machine learning-based controllers are essential to unlock greater energy efficiency and reliability, promoting sustainability. We present LC-Opt, a Sustainable Liquid Cooling (LC) benchmark environment, for reinforcement learning (RL) control strategies in energy-efficient liquid cooling of high-performance computing (HPC) systems. Built on the baseline of a high-fidelity digital twin of Oak Ridge National Lab's Frontier Supercomputer cooling system, LC-Opt provides detailed Modelica-based end-to-end models spanning site-level cooling towers to data center cabinets and server blade groups. RL agents optimize critical thermal controls like liquid supply temperature, flow rate, and granular valve actuation at the IT cabinet level, as well as cooling tower (CT) setpoints through a Gymnasium interface, with dynamic changes in workloads. This environment creates a multi-objective real-time optimization challenge balancing local thermal regulation and global energy efficiency, and also supports additional components like a heat recovery unit (HRU). We benchmark centralized and decentralized multi-agent RL approaches, demonstrate policy distillation into decision and regression trees for interpretable control, and explore LLM-based methods that explain control actions in natural language through an agentic mesh architecture designed to foster user trust and simplify system management. LC-Opt democratizes access to detailed, customizable liquid cooling models, enabling the ML community, operators, and vendors to develop sustainable data center liquid cooling control solutions.

Geometry optimization of atomic structures is a common and crucial task in computational chemistry and materials design. Following the learning to optimize paradigm, we propose a new multi-agent reinforcement learning method called Multi-Agent Crystal Structure optimization (MACS) to address the problem of periodic crystal structure optimization. MACS treats geometry optimization as a partially observable Markov game in which atoms are agents that adjust their positions to collectively discover a stable configuration. We train MACS across various compositions of reported crystalline materials to obtain a policy that successfully optimizes structures from the training compositions as well as structures of larger sizes and unseen compositions, confirming its excellent scalability and zero-shot transferability. We benchmark our approach against a broad range of state-of-the-art optimization methods and demonstrate that MACS optimizes periodic crystal structures significantly faster, with fewer energy calculations, and the lowest failure rate.

The tremendous potential has been demonstrated by large language models (LLMs) in intelligent decision-making problems, with unprecedented capabilities shown across diverse applications ranging from gaming AI systems to complex strategic planning frameworks. However, the StarCraft II platform, which has been widely adopted for validating decision-making algorithms in the past decade, has not yet provided substantial support for this emerging domain. To address issues that LLMs cannot interface with the hundreds of actions of the pysc2 backend and the lack of native support for multi-agent (MA) collaboration, we propose the LLM-PySC2 environment. This is the first environment that offers LLMs the complete pysc2 action space with sufficient multi-modal information and game Wiki knowledge. With an asynchronous query architecture, the environment efficiently interacts with LLMs that maintain a constant latency regardless of the scale of the agents' population. In the experiments, we evaluated LLMs' decision-making performance in both the macro-decision and micro-operation scenarios, with traditional StarCraft II Multi-Agent Challenge (SMAC) tasks and a series of new proposed. Results indicate that LLMs possess the potential to achieve victories in complex scenarios but cannot constantly generate correct decisions, especially in the recovered pysc2 action space and MA settings. Without task-relevant instructions, the pre-trained models suffer from issues such as hallucinations and inefficient collaboration. Our findings suggest that StarCraft II still challenges in the era of large models, revealing that there is a lot to do to develop an advanced LLM decision-making system, and the proposed LLM-PySC2 environment will support future development of LLM-based decision-making solutions.

With the rapid development of Large Vision Language Models, the focus of Graphical User Interface (GUI) agent tasks shifts from single-screen tasks to complex screen navigation challenges. However, real-world GUI environments, such as PC software and mobile Apps, are often complex and proprietary, making it difficult to obtain the comprehensive environment information needed for agent training and evaluation. This limitation hinders systematic investigation and benchmarking of agent navigation capabilities. To address this limitation, we introduce GUI Exploration Lab, a simulation environment engine for GUI agent navigation research that enables flexible definition and composition of screens, icons, and navigation graphs, while providing full access to environment information for comprehensive agent training and evaluation. Through extensive experiments, we find that supervised fine-tuning enables effective memorization of fundamental knowledge, serving as a crucial foundation for subsequent training. Building on this, single-turn reinforcement learning further enhances generalization to unseen scenarios. Finally, multi-turn reinforcement learning encourages the development of exploration strategies through interactive trial and error, leading to further improvements in screen navigation performance. We validate our methods on both static and interactive benchmarks, demonstrating that our findings generalize effectively to real-world scenarios. These findings demonstrate the advantages of reinforcement learning approaches in GUI navigation and offer practical guidance for building more capable and generalizable GUI agents.

What does it truly mean for a language model to “reason” strategically, and can scaling up alone guarantee intelligent, context-aware decisions? Strategic decision-making requires adaptive reasoning, where agents anticipate and respond to others’ actions under uncertainty. Yet, most evaluations of large language models (LLMs) for strategic decision-making often rely heavily on Nash Equilibrium (NE) benchmarks, overlook reasoning depth, and fail to reveal the mechanisms behind model behavior. To address this gap, we introduce a behavioral game-theoretic evaluation framework that disentangles intrinsic reasoning from contextual influence. Using this framework, we evaluate 22 state-of-the-art LLMs across diverse strategic scenarios. We find models like GPT-o3-mini, GPT-o1, and DeepSeek-R1 lead in reasoning depth. Through thinking chain analysis, we identify distinct reasoning styles—such as maximin or belief-based strategies—and show that longer reasoning chains do not consistently yield better decisions. Furthermore, embedding demographic personas reveals context-sensitive shifts: some models (e.g., GPT-4o, Claude-3-Opus) improve when assigned female identities, while others (e.g., Gemini 2.0) show diminished reasoning under minority sexuality personas. These findings underscore that technical sophistication alone is insufficient; alignment with ethical standards, human expectations, and situational nuance is essential for the responsible deployment of LLMs in interactive settings.

Photo retouching has become integral to contemporary visual storytelling, enabling users to capture aesthetics and express creativity. While professional tools such as Adobe Lightroom offer powerful capabilities, they demand substantial expertise and manual effort. In contrast, existing AI-based solutions provide automation but often suffer from limited adjustability and poor generalization, failing to meet diverse and personalized editing needs. To bridge this gap, we introduce JarvisArt, a multi-modal large language model (MLLM)-driven agent that understands user intent, mimics the reasoning process of professional artists, and intelligently coordinates over 200 retouching tools within Lightroom. JarvisArt undergoes a two-stage training process: an initial Chain-of-Thought supervised fine-tuning to establish basic reasoning and tool-use skills, followed by Group Relative Policy Optimization for Retouching (GRPO-R) to further enhance its decision-making and tool proficiency. We also propose the Agent-to-Lightroom Protocol to facilitate seamless integration with Lightroom. To evaluate performance, we develop MMArt-Bench, a novel benchmark constructed from real-world user edits. JarvisArt demonstrates user-friendly interaction, superior generalization, and fine-grained control over both global and local adjustments, paving a new avenue for intelligent photo retouching. Notably, it outperforms GPT-4o with a 60\% improvement in average pixel-level metrics on MMArt-Bench for content fidelity, while maintaining comparable instruction-following capabilities.

Prior studies on deception in language-based AI agents typically assess whether the agent produces a false statement about a topic, or makes a binary choice prompted by a goal, rather than allowing open-ended deceptive behavior to emerge in pursuit of a longer-term goal. To fix this, we introduce $\textit{Among Us}$, a sandbox social deception game where LLM-agents exhibit long-term, open-ended deception as a consequence of the game objectives. While most benchmarks saturate quickly, $\textit{Among Us}$ can be expected to last much longer, because it is a multi-player game far from equilibrium. Using the sandbox, we evaluate $18$ proprietary and open-weight LLMs and uncover a general trend: models trained with RL are comparatively much better at producing deception than detecting it. We evaluate the effectiveness of methods to detect lying and deception: logistic regression on the activations and sparse autoencoders (SAEs). We find that probes trained on a dataset of ``pretend you're a dishonest model: $\dots$'' generalize extremely well out-of-distribution, consistently obtaining AUROCs over 95% even when evaluated just on the deceptive statement, without the chain of thought. We also find two SAE features that work well at deception detection but are unable to steer the model to lie less. We hope our open-sourced sandbox, game logs, and probes serve to anticipate and mitigate deceptive behavior and capabilities in language-based agents.

Recent advances in computational pathology have led to the emergence of numerous foundation models. These models typically rely on general-purpose encoders with multi-instance learning for whole slide image (WSI) classification or apply multimodal approaches to generate reports directly from images. However, these models cannot emulate the diagnostic approach of pathologists, who systematically examine slides at low magnification to obtain an overview before progressively zooming in on suspicious regions to formulate comprehensive diagnoses. Instead, existing models directly output final diagnoses without revealing the underlying reasoning process. To address this gap, we introduce CPathAgent, an innovative agent-based approach that mimics pathologists' diagnostic workflow by autonomously navigating across WSI through zoom-in/out and move operations based on observed visual features, thereby generating substantially more transparent and interpretable diagnostic summaries. To achieve this, we develop a multi-stage training strategy that unifies patch-level, region-level, and WSI-level capabilities within a single model, which is essential for replicating how pathologists understand and reason across diverse image scales. Additionally, we construct PathMMU-HR², the first expert-validated benchmark for large region analysis. This represents a critical intermediate scale between patches and whole slides, reflecting a key clinical reality where pathologists typically examine several key large regions rather than entire slides at once. Extensive experiments demonstrate that CPathAgent consistently outperforms existing approaches across benchmarks at three different image scales, validating the effectiveness of our agent-based diagnostic approach and highlighting a promising direction for computational pathology.

In reinforcement learning, we typically refer to *unsupervised* pre-training when we aim to pre-train a policy without a priori access to the task specification, i.e., rewards, to be later employed for efficient learning of downstream tasks. In single-agent settings, the problem has been extensively studied and mostly understood. A popular approach casts the unsupervised objective as maximizing the *entropy* of the state distribution induced by the agent's policy, from which principles and methods follow. In contrast, little is known about state entropy maximization in multi-agent settings, which are ubiquitous in the real world. What are the pros and cons of alternative problem formulations in this setting? How hard is the problem in theory, how can we solve it in practice? In this paper, we address these questions by first characterizing those alternative formulations and highlighting how the problem, even when tractable in theory, is non-trivial in practice. Then, we present a scalable, decentralized, trust-region policy search algorithm to address the problem in practical settings. Finally, we provide numerical validations to both corroborate the theoretical findings and pave the way for unsupervised multi-agent reinforcement learning via state entropy maximization in challenging domains, showing that optimizing for a specific objective, namely *mixture entropy*, provides an excellent trade-off between tractability and performances.