arXiv:2609.19414v1 Announce Type: new Abstract: Goal-conditioned reinforcement learning aims to learn policies that reach specified goals, but remains challenging in offline settings with sparse rewards and long-horizon dependencies. In such settings, goal-completion information …
arXiv:2609.20098v1 Announce Type: new Abstract: Reliable temporal-difference targets are central to off-policy actor-critic learning. Direct value improvement refines the next-state target with alternative actions, but the reliability of this refinement depends on how candidate a…
arXiv:2609.20300v1 Announce Type: new Abstract: Current offline reinforcement learning (ORL) algorithms tend to overfit the training dataset and exhibit poor in-distribution generalization and robustness performance when deployed to real environments, thus compromising their effe…
arXiv cs.LG
TIER_1English(EN)·Martin Marek, Max Ryabinin·
arXiv:2609.20807v1 Announce Type: new Abstract: Reinforcement learning (RL) of large language models is notoriously sensitive to small differences between training and inference engines, often referred to as the training-inference mismatch (TIM). However, completely eliminating T…
arXiv:2609.20747v1 Announce Type: cross Abstract: Reinforcement learning constitutes a promising approach owing to its potential for superhuman performance and self-learned policies. However, its application to real-world autonomous driving remains scarce, particularly in unstruc…
arXiv:2609.20089v1 Announce Type: new Abstract: Self-evolving methods reduce the need for human-annotated trajectories by allowing tool-using agents to generate their own training data. Yet existing methods typically separate trajectory generation from evaluation, relying on stat…
arXiv cs.AI
TIER_1English(EN)·Nikita Khomich, Leopold Hermansson, Ido Hakimi·
arXiv:2609.20004v1 Announce Type: cross Abstract: Reward-based reinforcement learning for language models, exemplified by Group Relative Policy Optimization (GRPO), collapses an entire stochastic trajectory into a single scalar reward. This is clean and scalable, but it explores …
arXiv cs.AI
TIER_1English(EN)·Yilang Liu, Haoxiang You, Qian Wang, Daniel Rakita, Ian Abraham·
arXiv:2609.20575v1 Announce Type: cross Abstract: Learning visual policies for locomotion and manipulation requires coordinating contact with the environment and can incur substantial computation and GPU memory costs. First-order policy gradients (FoPG) reduce training cost throu…
arXiv cs.AI
TIER_1English(EN)·Jaemoo Choi, Yuchen Zhu, Wei Guo, Petr Molodyk, Bo Yuan, Jinbin Bai, Yi Xin, Molei Tao, Yongxin Chen·
arXiv:2602.04663v3 Announce Type: replace-cross Abstract: Reinforcement learning has been widely applied to diffusion and flow models for visual tasks such as text-to-image generation. However, these tasks remain challenging because diffusion models have intractable likelihoods, …
arXiv cs.AI
TIER_1English(EN)·Shabir Ahmad Sofi, Bisma Majid, Mir Mohammad Yousuf·
arXiv:2609.18219v1 Announce Type: cross Abstract: Quantum Reinforcement Learning (QRL) represents policies as variational quantum circuits (VQCs), making it attractive for combinatorial optimization such as the Capacitated Vehicle Routing Problem (CVRP). On noisy intermediate-sca…
arXiv cs.LG
TIER_1English(EN)·Giray \"On\"ur, Azita Dabiri, Bart De Schutter·
arXiv:2609.17697v1 Announce Type: new Abstract: Integrated deep reinforcement learning (DRL) and model predictive control (MPC) methods are increasingly used to control autonomous systems by combining their complementary capabilities. DRL learns control policies through interacti…
arXiv:2609.18964v1 Announce Type: new Abstract: Federated Reinforcement Learning (FRL) enables collaborative policy learning across distributed agents with heterogeneous environments. While recent methods based on variance reduction, divergence penalization, and momentum optimiza…
arXiv cs.LG
TIER_1English(EN)·Everest Yang, Skye Thompson, George D. Konidaris·
arXiv:2609.18167v1 Announce Type: cross Abstract: Adapting to changes in robot dynamics requires learning from new data without discarding experience that may still be useful. In continual model-based reinforcement learning (RL), replay collected before a dynamics change can slow…
arXiv:2604.14765v2 Announce Type: replace Abstract: We present a geometric framework for Reinforcement Learning (RL) that views policies as maps into the Wasserstein space of action probabilities. First, we define a Riemannian structure induced by stationary distributions, provin…
arXiv cs.AI
TIER_1English(EN)·Trishit Mondal, Irina Tezaur, Anthony Gruber·
arXiv:2609.17837v1 Announce Type: cross Abstract: Hybrid domain decomposition methods provide a flexible framework for coupling full order models (FOMs) and reduced order models (ROMs), but typically assume the model assigned to each subdomain is fixed throughout a simulation. Th…
arXiv cs.AI
TIER_1English(EN)·Bernd Frauenknecht, Emma Cramer, Artur Eisele, Paul Kruse, Lukas Kesper, Jonas Hertrampf, Ramil Sabirov, Jyotirmaya Patra, Johannes Berger, Paul Brunzema, Friedrich Solowjow, Sebastian Trimpe·
arXiv:2609.19074v1 Announce Type: cross Abstract: Reinforcement learning (RL) is an exciting concept as well as a remarkable success story worth sharing. However, RL builds on rather complex interactions between different objects that play out over several cycles. Such dynamics a…
arXiv cs.AI
TIER_1English(EN)·Tuan Dam, Kishan Panaganti, Brahim Driss, Adam Wierman·
arXiv:2609.18599v1 Announce Type: cross Abstract: Monte Carlo Tree Search (MCTS) is a powerful framework for solving complex decision-making problems, yet it often relies on the assumption that the simulator and the real-world dynamics are identical. Although this assumption help…
arXiv:2609.18587v1 Announce Type: cross Abstract: Test-time reinforcement learning (TTRL) enables models to improve their reasoning without relying on labeled training data, but existing approaches typically optimize a large fraction of the model parameters. This raises a natural…
arXiv cs.AI
TIER_1English(EN)·Gabriel Nova, Stephane Hess, Sander Van Cranenburgh·
arXiv:2609.18441v1 Announce Type: cross Abstract: Discrete choice model specification is a time-consuming task in which modellers often specify and estimate multiple models while balancing goodness-of-fit, parsimony, and behavioural plausibility. We present Delphos, a multitask r…
Reinforcement learning (RL) is an exciting concept as well as a remarkable success story worth sharing. However, RL builds on rather complex interactions between different objects that play out over several cycles. Such dynamics are often best explained with an easily accessible …
arXiv:2609.17115v1 Announce Type: cross Abstract: Vision-language-action (VLA) systems already bring together two valuable resources for robot learning: rich visual representations and demonstrations of successful task execution. Intrinsic Robot Rewarding (IRR) proposes to use th…
arXiv:2609.17325v1 Announce Type: new Abstract: Biological cells can be viewed as individual, interacting agents whose collective dynamics give rise to adaptive behaviour at multiple levels of organisation, from individual cells through tissues to whole multicellular organisms. I…
arXiv cs.LG
TIER_1English(EN)·Saki Omi, Hyo-Sang Shin, Namhoon Cho, Antonios Tsourdos, Miguel A. Olivares-Mendez·
arXiv:2307.15931v3 Announce Type: replace Abstract: Reinforcement learning (RL) performs well in continuous-control tasks, but evolving hidden disturbances create partial observability: the agent must infer decision-relevant latent dynamics from interaction history. This study in…
arXiv cs.LG
TIER_1English(EN)·Pablo Rodriguez-Grasa, Sofiene Jerbi, Mikel Sanz, Ryan Sweke·
arXiv:2609.16266v1 Announce Type: cross Abstract: In recent years, the utility of parameterized quantum circuits as function approximators has been widely studied. In the context of reinforcement learning, this approach has led to variational quantum algorithms such as quantum Q-…
arXiv:2609.14896v1 Announce Type: cross Abstract: A notable byproduct of LLM alignment training is mode collapse: the progressive loss of output diversity that narrows a model's expressivity at inference time. This degradation is especially limiting for applications requiring ope…
arXiv:2609.13545v1 Announce Type: cross Abstract: Learning-based adaptive control of robotic manipulators with non-observable friction memory has been addressed by attention- based meta-controllers whose number of attention heads is fixed before training and is tuned by costly of…
arXiv cs.LG
TIER_1English(EN)·Amal Dev Haridevan, Junjie Kang, Jinjun Shan·
arXiv:2609.15915v1 Announce Type: new Abstract: Meta-reinforcement learning (meta-RL) enables agents to adapt to unseen tasks with limited experience. Despite its promise, the application of meta-RL in real-world tasks is hindered by safety requirements, which have been underexpl…
arXiv cs.LG
TIER_1English(EN)·Paul Stahlhofen, Luca Hermes, Tim Kochs, Markus Vieth, Barbara Hammer·
arXiv:2609.15297v1 Announce Type: new Abstract: In order to make Reinforcement Learning algorithms applicable in real world scenarios, safety must be ensured even under adverse operating conditions. In this work, we consider the challenge of adversarial feature missingness: a sce…
arXiv cs.LG
TIER_1English(EN)·Lingheng Du, Yiming Tang, Xufeng Duan, Dianbo Liu·
arXiv:2609.15064v1 Announce Type: new Abstract: Reinforcement learning (RL) is widely utilized in large language model training to improve targeted capabilities, yet how RL reshapes a model remains poorly understood. Prior attempts to explain how RL works largely offer behavioral…
arXiv:2609.14327v1 Announce Type: new Abstract: Variance penalization is a principled approach to risk-sensitive reinforcement learning (RL) that explicitly trades expected return for policy stability. Existing methods require a dedicated second critic to estimate return variance…
arXiv cs.AI
TIER_1English(EN)·Aadyot Bhatnagar, Peter M{\o}rch Groth, Sebastian Ibarraran, Ali Madani·
arXiv:2604.13175v2 Announce Type: replace-cross Abstract: Large language models can be aligned with human preferences through offline reinforcement learning (RL) on small labeled datasets. While single-objective alignment is well-studied, many real-world applications demand the s…
arXiv cs.AI
TIER_1English(EN)·Yuning Wu, Ke Wang, Haoran Liu, Chaoqun Jia, Devin Chen, Kai Wei·
arXiv:2603.11321v3 Announce Type: replace-cross Abstract: Reinforcement Learning with Verifiable Rewards improves reasoning in large language models, yet on-policy learning often suffers from cold-start challenges in sparse-reward settings. Recent mixed-policy approaches address …
arXiv:2609.14647v1 Announce Type: cross Abstract: Robots, and humanoid robots in particular, are increasingly competent at individual behaviors, each obtained by training a specialized controller. A specialized skill is quick to train, converges reliably because the problem it fa…
arXiv cs.AI
TIER_1English(EN)·Hongliang Wei (Harbin Institute of Technology, Alibaba Cloud), Xiaobing Tu (Alibaba Cloud), Yinggui Wang (Alibaba Cloud), Zhengxi Liu (Alibaba Cloud), Rongkun Xue (Alibaba Cloud), Jinkui Ren (Alibaba Cloud), Xiantao Zhang (Alibaba Cloud), Debin Zhao (Har…·
arXiv:2609.13739v1 Announce Type: cross Abstract: Language-model agents are increasingly deployed through diverse harnesses that differ in system prompts, tool schemas, control loops, and trajectory formats. The same model can perform unevenly across these interfaces, making robu…
arXiv:2609.13425v1 Announce Type: cross Abstract: Training diffusion models with multiple rewards requires distinguishing user preference from reward informativeness. User preference determines how much each reward should contribute to the overall objective; reward informativenes…
arXiv:2609.13259v1 Announce Type: cross Abstract: Virtual Try-On (VTON) aims to dress a person with the reference garment, producing visually reasonable results aligned with human preferences. Turning this preference-oriented goal into an actionable objective relies on a scoring …
arXiv:2609.13194v1 Announce Type: cross Abstract: A large literature builds physical structure into learned dynamics on the premise that models respecting the underlying physics predict better. We test that premise using exact polynomial invariants recovered from trajectories and…
arXiv cs.AI
TIER_1English(EN)·Lindsay Spoor, Aske Plaat, Thomas Moerland·
arXiv:2609.15315v1 Announce Type: new Abstract: Safe reinforcement learning (RL) is commonly formalized as a Constrained Markov Decision Process (CMDP), in which an agent maximizes expected reward while keeping its expected cumulative cost below a specified safety bound. Existing…
arXiv:2609.12424v1 Announce Type: new Abstract: Reinforcement learning is now the standard way to train large language model agents on long-horizon tasks, where dozens of interdependent actions precede a single sparse reward. Critic-free, group-relative methods such as GRPO suit …
arXiv:2609.12014v1 Announce Type: new Abstract: Safe offline reinforcement learning assumes a cost function on every transition. We ask what remains possible when safety can be judged only by comparing short clips and occasionally asking whether an episode exceeded its budget. Ce…
arXiv:2609.13058v1 Announce Type: new Abstract: Reinforcement learning (RL) has become central to post-training of large language models. Recent advances in RL for Mixture-of-Experts (MoE) models have primarily focused on improving optimization stability and training efficiency, …
arXiv:2609.12896v1 Announce Type: new Abstract: LLM-based agents rely on heterogeneous interaction capabilities to accomplish complex tasks. Existing approaches often distribute these capabilities across multiple LoRA adapters, which increases adapter storage requirements and int…
arXiv:2609.13035v1 Announce Type: new Abstract: Symmetries play a central role in reducing the complexity of reinforcement learning problems, yet most existing approaches rely on fixed group actions or predefined state abstractions. Classical reinforcement learning algorithms typ…
arXiv:2609.13050v1 Announce Type: new Abstract: Regularization-based methods have become a standard approach for training Deep Reinforcement Learning policies against adversarial input perturbations. In this paper, we unify these methods by deriving new upper bounds on the perfor…
arXiv cs.LG
TIER_1English(EN)·Fernando Palafox, David Fridovich-Keil·
arXiv:2609.12278v1 Announce Type: new Abstract: World models let agents plan by predicting the consequences of their actions, but changes in the environment can make them inaccurate. We study the problem of adapting a world model to an unknown test-time environment, drawn from a …
A notable byproduct of LLM alignment training is mode collapse: the progressive loss of output diversity that narrows a model's expressivity at inference time. This degradation is especially limiting for applications requiring open-ended exploration and pluralistic perspectives, …
arXiv:2609.11061v1 Announce Type: new Abstract: Tree-structured rollouts give critic-free reinforcement learning with verifiable rewards (RLVR) step-level credit: fork a chain at an intermediate point, and sibling outcome differences estimate step value. Each fork adds sampling c…
arXiv cs.AI
TIER_1English(EN)·Martin Klissarov, Akhil Bagaria, Ziyan Luo, George Konidaris, Doina Precup, Marlos C. Machado·
arXiv:2506.14045v2 Announce Type: replace Abstract: Developing agents capable of exploring, planning and learning in complex open-ended environments is a grand challenge in artificial intelligence (AI). Hierarchical reinforcement learning (HRL) offers a promising solution to this…
arXiv:2306.13732v2 Announce Type: replace Abstract: We study a class of reinforcement learning (RL) tasks where the objective of the agent is to accomplish temporally extended goals. In this setting, a common approach is to represent the tasks as deterministic finite automata (DF…
arXiv:2609.11042v1 Announce Type: new Abstract: Agent usage is shifting toward long-horizon tasks such as coding and scientific discovery, among which terminal tasks are especially important. We introduce T1, a Mixture-of-Experts model of 122B total trained with reinforcement lea…
arXiv:2609.10866v1 Announce Type: new Abstract: Reinforcement learning (RL) agents deployed in real-world environments are often vulnerable to adversarial perturbations in state observations, creating risks in safety-critical applications. Certification methods can improve robust…
arXiv cs.AI
TIER_1English(EN)·Kaiyan Zhao, Yiming Wang, Yuyang Chen, Yan Li, Leong Hou U, Xiaoguang Niu·
arXiv:2410.20487v5 Announce Type: replace-cross Abstract: Experience replay is widely used to improve learning efficiency in reinforcement learning by leveraging past experiences. However, existing experience replay methods, whether based on uniform or prioritized sampling, often…
arXiv:2609.10315v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) has advanced language-model reasoning in domains such as mathematics and code, where objective answers are inexpensive to check. Diagnostic reasoning over complex data lacks this…
arXiv:2512.10835v2 Announce Type: replace Abstract: This paper introduces a reinforcement learning framework that enables controllable and diverse player behaviors without relying on human gameplay data. Existing approaches often require large-scale player trajectories, train sep…
ESRL improves reinforcement learning for mixture-of-experts language models by exploring expert routing with anchored high-confidence experts, entropy-adaptive perturbation, and path replay to boost rollout diversity and training performance.
Tree-structured rollouts give critic-free reinforcement learning with verifiable rewards (RLVR) step-level credit: fork a chain at an intermediate point, and sibling outcome differences estimate step value. Each fork adds sampling cost, so realistic budgets typically allow only a…
arXiv cs.LG
TIER_1English(EN)·Pablo Boitel, Kun Zhang·
arXiv:2609.05822v1 Announce Type: new Abstract: Deploying advanced HVAC (Heating, Ventilation and Air Conditioning) controllers at scale remains difficult because performance often depends on accurate building models or per-site retuning. We propose NOMAD-RL (Neural Online Meta-A…
arXiv cs.AI
TIER_1English(EN)·Gangyi Zhang, Junjie Meng, Letian Zhang, Wei Wu, Yang Zheng, Dong Wang, Yang Liu, Guanjun Jiang, Chongming Gao·
arXiv:2609.07247v1 Announce Type: new Abstract: Scaling the interaction horizon-the maximum number of environment interactions per episode-improves LLM agents on long-horizon tasks, and curriculum-based methods that progressively expand the horizon outperform fixed-horizon altern…
arXiv cs.AI
TIER_1English(EN)·Daniel Palenicek, Mikael Henaff, Scott Fujimoto, Koustuv Sinha·
arXiv:2609.06421v1 Announce Type: cross Abstract: Batch normalization (BN) substantially improves sample efficiency in continuous-control actor-critic methods such as CrossQ, yet recent studies report performance degradation in discrete-action value learning on Atari. These failu…
arXiv cs.AI
TIER_1English(EN)·Mahmoud Selim, Cristina Cipriani, Karl H. Johansson·
arXiv:2609.06882v1 Announce Type: cross Abstract: Diffusion policies offer a powerful and expressive parameterization for continuous control. Yet, their integration with reinforcement learning remains conceptually and algorithmically challenging. In this work, we address this gap…
arXiv cs.AI
TIER_1English(EN)·Yacine El Yamani, Hanna Krasowski, Elena Vanneaux·
arXiv:2609.07618v1 Announce Type: cross Abstract: Environments are increasingly populated by multiple robots performing independent tasks with limited prior knowledge of each other. Deploying such multi-agent systems presents significant challenges. Specifically, shifts in deploy…
arXiv:2609.08642v1 Announce Type: cross Abstract: Exploration in reinforcement learning (RL) remains a fundamental challenge. Recent goal-conditioned RL strategies (which select goals to encourage broader state coverage) have shown promising results, but none scores a goal by nov…
arXiv cs.AI
TIER_1English(EN)·Olivier Lepel, Anas Barakat·
arXiv:2410.02605v5 Announce Type: replace-cross Abstract: We derive a policy gradient theorem for Cumulative Prospect Theory (CPT) objectives in finite-horizon Reinforcement Learning (RL), generalizing the standard policy gradient theorem and encompassing distortion-based risk ob…
arXiv cs.CL
TIER_1English(EN)·Zikang Shan, Han Zhong, Liwei Wang, Li Zhao·
arXiv:2604.10701v2 Announce Type: replace-cross Abstract: Credit assignment is a central challenge in reinforcement learning (RL). Classical actor-critic methods address this challenge through fine-grained advantage estimation based on a learned value function. However, learned v…
arXiv:2609.05650v1 Announce Type: new Abstract: We propose a reinforcement learning framework in which exploration is driven by intrinsic curiosity, designed for scenarios where environments are non-stationary and rewards are sparse, delayed, uninformative, or absent. In our mode…
arXiv:2609.06186v1 Announce Type: new Abstract: Prioritized Sweeping (PS) accelerates model-based reinforcement learning by selecting backups according to Bellman residual magnitude. In nonstationary reward settings, however, the canonical priority score is shortsighted: after a …
arXiv:2609.06467v1 Announce Type: new Abstract: In performative reinforcement learning the deployed policy shapes the environment that generates the learner's future data, and the natural solution concept is a performatively stable policy that is optimal in the environment it ind…
arXiv cs.LG
TIER_1English(EN)·Zihan Lin, Xiaohan Wang, Jie Cao, Jiajun Chai, Guojun Yin, Wei Lin, Ran He·
arXiv:2609.06469v1 Announce Type: new Abstract: Reinforcement learning (RL) across multiple domains can broaden the reasoning capabilities of large language models (LLMs), yet joint training often degrades individual-domain performance and can destabilize optimization. Existing w…
arXiv cs.LG
TIER_1English(EN)·Jan Corazza, Daniil Kaminskyi, Simon Lutz, Patrick Nossol, Hadi Partovi Aria, Zhe Xu, Daniel Neider·
arXiv:2609.07461v1 Announce Type: new Abstract: We consider reinforcement learning in environments with dynamics that undergo an irreversible phase transition governed by a hidden temporal pattern. The agent observes the base state but cannot observe the phase directly. We formal…
arXiv:2609.08080v1 Announce Type: new Abstract: Ensuring safety in reinforcement learning under nonstationarity requires anticipating changes in risk before they lead to unsafe behavior. Existing approaches typically rely on safety constraints defined at design time or updated re…
arXiv cs.LG
TIER_1English(EN)·Ryan Truong, Lance Ying, Samuel J. Gershman, Kazuki Irie·
arXiv:2609.09059v1 Announce Type: new Abstract: While many video-game environments (VGEs) have played crucial roles in advancing reinforcement learning (RL), developing novel VGEs or modifying existing ones to support new features, has been a laborious process requiring extensive…
arXiv cs.LG
TIER_1English(EN)·Mohammadreza Nakheai, Aidan Scannell, Kevin Luck, Joni Pajarinen·
arXiv:2603.02935v2 Announce Type: replace Abstract: Offline meta-reinforcement learning seeks to learn a policy that generalizes to new related tasks online. Context-based methods infer a task representation from transition histories, yet learning an effective task representation…
T1 is a 122B Mixture-of-Experts model trained with reinforcement learning to execute long-horizon terminal tasks in a cloud sandbox, achieving state-of-the-art results through stable actor-critic optimization and out-of-distribution training.
Finding new physics (NP) is the most important problem in particle physics today. Studying ``anomalies'', i.e., measurements of low-energy observables whose values disagree with the predictions of the Standard Model (SM), is a powerful search strategy. The SM Effective Field Theo…
Exploration in reinforcement learning (RL) remains a fundamental challenge. Recent goal-conditioned RL strategies (which select goals to encourage broader state coverage) have shown promising results, but none scores a goal by novelty and reachability jointly: the two signals are…
Environments are increasingly populated by multiple robots performing independent tasks with limited prior knowledge of each other. Deploying such multi-agent systems presents significant challenges. Specifically, shifts in deployment states compared to training data can lead to …
arXiv:2609.05298v1 Announce Type: cross Abstract: Cooperative multi-agent reinforcement learning (MARL) systems rely on past experience for learning coordinated behaviour, but this experience may become unreliable if the environment or task objective changes during training. In s…
arXiv:2602.05765v3 Announce Type: replace Abstract: Reinforcement learning (RL) has emerged as a critical paradigm for post-training Vision-Language-Action (VLA) models, enabling embodied agents to adapt and improve through environmental interaction. However, existing RL framewor…
arXiv cs.LG
TIER_1English(EN)·Abdulaziz Almuzairee, Henrik I. Christensen·
arXiv:2602.21203v2 Announce Type: replace-cross Abstract: Visual reinforcement learning is appealing for robotics but expensive. Off-policy methods are sample-efficient yet slow while on-policy methods parallelize well but waste samples. Recent work has shown that off-policy meth…
arXiv:2609.04865v1 Announce Type: new Abstract: Skill libraries improve the sample efficiency of agentic reinforcement learning (RL) by enabling large language model (LLM) agents to reuse procedural knowledge. Yet existing paradigms exhibit structural shortcomings: they either de…
Cooperative multi-agent reinforcement learning (MARL) systems rely on past experience for learning coordinated behaviour, but this experience may become unreliable if the environment or task objective changes during training. In such cases, agents first need a way to recognize th…
arXiv:2609.02931v1 Announce Type: cross Abstract: Scripted and rule-based non-player characters (NPCs) in combat video games often exhibit predictable behaviors that experienced players can exploit, while reinforcement learning (RL) agents typically retain a fixed policy after tr…
arXiv cs.LG
TIER_1English(EN)·Angel Y. He, David Parker·
arXiv:2609.04189v1 Announce Type: new Abstract: We introduce the first Probably Approximately Correct (PAC) learning framework for general-sum concurrent stochastic games (CSGs) with transition uncertainty, while addressing the challenge of Nash equilibrium (NE) existence. Our al…
arXiv cs.LG
TIER_1English(EN)·Charlotte Beylier, Hannah Selder, Arthur Fleig, Simon M. Hofmann, Nico Scherf·
arXiv:2511.20591v3 Announce Type: replace Abstract: The emergence and evolution of feature reliance in deep reinforcement learning agents remain poorly understood. Here, we introduce a methodological framework for analyzing the learning process through quantitative analysis of sa…
arXiv:2609.03842v1 Announce Type: new Abstract: Offline reinforcement learning (RL) must reconcile two competing requirements: policy updates should stay near dataset-supported actions to keep value estimates reliable, yet meaningful gains often require moving beyond the behavior…
arXiv:2502.18407v2 Announce Type: replace-cross Abstract: Existing LLM-based agents have achieved strong performance on held-in tasks, but their generalizability to unseen tasks remains poor. Hence, some recent work focus on fine-tuning the policy model with more diverse tasks to…
arXiv cs.AI
TIER_1English(EN)·Oussama Hidaoui, Omer Ebead, Ulrich Armel Mbou Sob, Siddarth Singh, Juan Claude Formanek, Felix Chalumeau, Omayma Mahjoub, Sasha Abramowitz, Ruan John de Kock, Wiem Khlifi, Louay Ben Nessir, Simon Verster Du Toit, Daniel Rajaonarivonivelomanantsoa, Asim …·
arXiv:2609.03667v1 Announce Type: cross Abstract: Generalising to unseen tasks remains a fundamental challenge in offline multi-agent reinforcement learning (MARL). In this work, we present a principled analysis of zero-shot task generalisation in the offline setting and conduct …
arXiv cs.AI
TIER_1English(EN)·Vincenzo Norman Vitale, Mohammad Solki, Antonia Maria Tulino, Andreas F. Molisch, Jaime Llorca·
arXiv:2609.03590v1 Announce Type: cross Abstract: Timely delivery of delay-sensitive information over dynamic, heterogeneous networks is essential for NextG interactive applications, yet providing strict End-to-End (E2E) peak latency guarantees remains an open challenge. Two obst…
We introduce the first Probably Approximately Correct (PAC) learning framework for general-sum concurrent stochastic games (CSGs) with transition uncertainty, while addressing the challenge of Nash equilibrium (NE) existence. Our algorithm maintains data-driven $L^1$ confidence s…
We introduce the first Probably Approximately Correct (PAC) learning framework for general-sum concurrent stochastic games (CSGs) with transition uncertainty, while addressing the challenge of Nash equilibrium (NE) existence. Our algorithm maintains data-driven $L^1$ confidence s…
Timely delivery of delay-sensitive information over dynamic, heterogeneous networks is essential for NextG interactive applications, yet providing strict End-to-End (E2E) peak latency guarantees remains an open challenge. Two obstacles limit the adoption of learning-based network…
arXiv:2410.02898v3 Announce Type: replace-cross Abstract: Reach-Avoid-Stay (RAS) tasks are essential in applications where systems must safely reach a target set and remain within it under all bounded disturbances. Existing approaches either struggle to compute the maximal robust…
arXiv cs.LG
TIER_1English(EN)·Martin van der Schelling, Deepesh Toshniwal, Miguel A. Bessa·
arXiv:2609.01811v1 Announce Type: cross Abstract: No single optimization method is uniformly best for all problems, and the most suitable optimizer choice can change during a run. Existing approaches that change optimizer during execution typically predetermine part of the strate…
arXiv cs.LG
TIER_1English(EN)·Pritthijit Nath, Sebastian Schemm, Peter Haynes, Emily Shuckburgh, Mark Webb·
arXiv:2609.02566v1 Announce Type: new Abstract: Machine-learnt corrections can complement numerical weather prediction only if they adapt to the evolving model state while preserving dynamical consistency and numerical stability. To test this within a global forecasting model, we…
Machine-learnt corrections can complement numerical weather prediction only if they adapt to the evolving model state while preserving dynamical consistency and numerical stability. To test this within a global forecasting model, we couple the Met Office (UKMO) Unified Model (UM)…
arXiv cs.LG
TIER_1English(EN)·Tom\'a\v{s} Hole\v{c}ek, Viliam Lis\'y·
arXiv:2609.01549v1 Announce Type: new Abstract: Model-based reinforcement learning (MBRL) has achieved remarkable results in single-agent domains, yet its extension to competitive imperfect information games (IIGs) remains underexplored. In multi-agent settings, opponent-induced …
arXiv cs.CL
TIER_1English(EN)·Yu Yuan, Yaoyou Fan, Lili Zhao, Guangting Zheng, Kai Zhang, Lu Pan, Ke Zeng, Qi Liu·
arXiv:2609.00213v1 Announce Type: new Abstract: Reinforcement learning fine-tuning of large language models increasingly adopts multiple reward dimensions, including verifiable rules, task-specific evaluators, and learned reward models, to provide richer supervision across divers…
arXiv:2609.01597v1 Announce Type: cross Abstract: Natural language is emerging as a primary feedback channel for improving language agents, capable of conveying intent, preferences, and causal structure in forms interpretable by both humans and modern language models. We call thi…
arXiv cs.AI
TIER_1English(EN)·Liming Pu, Xiaoxia Li, Yifu Liu, Teng Cao, Bin Yang·
arXiv:2609.01245v1 Announce Type: cross Abstract: Reinforcement learning is a natural way to post-train LLM agents for long-horizon interactive tasks judged only by end-of-task verification, yet a shared belief holds that outcome-only RL soon hits a ceiling on small open models. …
arXiv cs.AI
TIER_1English(EN)·Asrin Efe Yorulmaz, Ugur Aydin, Tamer Basar·
arXiv:2609.00504v1 Announce Type: cross Abstract: In this work, we study radically uncoupled learning in discounted general-sum Markov games. Assuming ``$\mathsf{ETH}$ for $\mathsf{PPAD}$", we show that, for every fixed discount factor, there is no polynomial-time algorithm for c…
arXiv cs.LG
TIER_1English(EN)·Baha Zarrouki, Arslan Thobani, Jasper Hoffmann, Mattia Piccinini, Rudolf Reiter, Felix Jahncke, S\'ebastien Gros, Davide Scaramuzza, Johannes Betz·
arXiv:2609.01061v1 Announce Type: cross Abstract: In Model Predictive Control (MPC), cost-function weights shape closed-loop behavior, yet changing conditions often make fixed parametrizations suboptimal and motivate context-dependent online adaptation. Learning such policies is …
arXiv cs.NE (Neural & Evolutionary)
TIER_1English(EN)·Miguel A. Bessa·
No single optimization method is uniformly best for all problems, and the most suitable optimizer choice can change during a run. Existing approaches that change optimizer during execution typically predetermine part of the strategy: the portfolio is restricted to one algorithm c…
Reinforcement learning for user-centric agents is limited by the cost, latency, and risk of collecting online feedback, as well as by the lack of counterfactual comparisons under the same user state. In this paper, we propose World Model-Guided Reinforcement Learning via counterf…
arXiv cs.LG
TIER_1English(EN)·Sheeraja Rajakrishnan, Alexander G. Ororbia, Travis Desell, Daniel E. Krutz·
arXiv:2608.29860v1 Announce Type: new Abstract: Uncertainty estimation provides promising capabilities for reinforcement learning (RL) agents. Notably, estimating uncertainty can reduce the training time and enable agents to obtain greater rewards over time by exploiting informat…
arXiv cs.LG
TIER_1English(EN)·Bastien Dubail, Stefan Stojanovic, Alexandre Prouti\`ere·
arXiv:2509.05193v3 Announce Type: replace Abstract: Low-rank structure is a common implicit assumption in many modern reinforcement learning (RL) algorithms. For instance, reward-free and goal-conditioned RL methods often presume that the successor measure admits a low-rank repre…
Scripted and rule-based non-player characters (NPCs) in combat video games often exhibit predictable behaviors that experienced players can exploit, while reinforcement learning (RL) agents typically retain a fixed policy after training and cannot readily adapt their strategy to …
Modern multi-agent systems are increasingly deployed at scale over large populations of agents in settings such as ad-auctions, traffic routing, and recommendation systems. The dominant approach in such settings is to optimize each agent's policy independently, treating the other…
arXiv:2609.20389v1 Announce Type: new Abstract: Offline policy evaluation (OPE) is crucial in high-stakes reinforcement learning applications, where new policies must be assessed reliably before deployment. In such settings, point estimates alone are insufficient; principled unce…
arXiv:2609.19418v1 Announce Type: cross Abstract: In evidence-based policymaking, typically one experimental sample is observed, then a learned policy recommendation is implemented at scale. Policies learned from the experimental data can perform well in expected welfare, yet ran…
arXiv:2609.19830v1 Announce Type: cross Abstract: Reinforcement learning for LLM agents involves two distinct optimization di- mensions: how environment feedback is exploited within a trajectory, and how complete trajectories are aggregated across a batch. We formulate these dime…
arXiv:2609.17296v1 Announce Type: cross Abstract: Policy learning aims to determine who should be treated based on individual characteristics. In high-stakes settings such as medicine and public policy where safety is a central concern, improving the average outcomes alone may no…
arXiv stat.ML
TIER_1English(EN)·Corentin Pla, Hugo Richard, Marc Abeille, Vianney Perchet·
arXiv:2609.11807v1 Announce Type: new Abstract: We study reinforcement learning (RL) with transition look-ahead, where the agent may observe which states would be visited upon playing any sequence of $\ell$ actions before deciding its course of action. Although look-ahead can sub…
arXiv:2512.06956v2 Announce Type: replace Abstract: Inverse reinforcement learning aims to infer the reward function that explains expert behavior observed through trajectories of state--action pairs. A long-standing difficulty in classical IRL is the non-uniqueness of the recove…
arXiv stat.ML
TIER_1English(EN)·Nan Lu, Ethan Lee, James M. Robins, David Simchi-Levi, Junwei Lu·
arXiv:2609.09981v1 Announce Type: new Abstract: We study offline inference for the optimal value in reinforcement learning. Two new nuisances are derived as fixed points of a self-induced Bellman equation, in which we approximate the maximum Bellman operator by its softmax corres…
arXiv cs.CV
TIER_1English(EN)·Jai Bardhan, Patrik Drozdik, Josef Sivic, Vladimir Petrik·
arXiv:2603.25685v2 Announce Type: replace-cross Abstract: Action-conditioned robot world models generate future video frames of the manipulated scene given a robot action sequence, offering a promising alternative for simulating tasks that are difficult to model with traditional …
arXiv:2609.02928v1 Announce Type: cross Abstract: Modern multi-agent systems are increasingly deployed at scale over large populations of agents in settings such as ad-auctions, traffic routing, and recommendation systems. The dominant approach in such settings is to optimize eac…
arXiv stat.ML
TIER_1English(EN)·Shrinivas Ramasubramanian, Daman Arora, Fahim Tajwar, Guanning Zeng, Qingyang Wu, Zhongzhu Zhou, Chenfeng Xu, Haiwen Feng, Yuda Song, Aarti Singh, Ruslan Salakhutdinov, J. Andrew Bagnell, Jeff Schneider, Andrea Zanette·
arXiv:2609.02987v1 Announce Type: cross Abstract: Reinforcement learning typically optimizes average reward. For generative policies, the average can hide an important distinction: two policies can achieve the same mean reward while having very different chances of producing a ra…
🤖 【Hugging Face Papers】T1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks Agent usage is shifting toward long-horizon tasks such as coding and scientific discovery, among which terminal tasks ar... # AI # TechNews # MachineLearning 🔗 https:// huggingface.co/papers/2…
<p>Large language models don't become aligned simply because they become larger. They become useful when their capabilities are shaped toward what humans actually prefer.</p> <p>A language model can generate grammatically correct text, write code, summarize documents, answer ques…
<p><em>Researchers have introduced Perturbed Parameter Policy Optimization (3PO), a new method that shifts reinforcement learning exploration from output distributions to the model's parameter space. By sampling diverse policies directly, 3PO reduces training instability and impr…
