PyTorchKR Papers

최신 논문

총 12,283편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2504.05185공식 코드 · Papers with Code

Concise Reasoning via Reinforcement Learning

Despite significant advancements in large language models (LLMs), a major drawback of reasoning models is their enormous token usage, which increases computational cost, resource requirements, and response time. In this work, we revisit the core principles of reinforcement learning (RL) and, through mathematical analysis, demonstrate that the tendency to generate lengthy responses arises inherently from RL-based optimization during training. This finding questions the prevailing assumption that longer responses inherently improve reasoning accuracy. Instead, we uncover a natural correlation between conciseness and accuracy that has been largely overlooked. We show that introducing a secondary phase of RL training, using a very small set of problems, can significantly reduce chains of thought while maintaining or even enhancing accuracy. Additionally, we demonstrate that, while GRPO shares some interesting properties of PPO, it suffers from collapse modes, which limit its reliability for concise reasoning. Finally, we validate our conclusions through extensive experimental results.

arXiv:2504.07257

Better Decisions through the Right Causal World Model

Reinforcement learning (RL) agents have shown remarkable performances in various environments, where they can discover effective policies directly from sensory inputs. However, these agents often exploit spurious correlations in the training data, resulting in brittle behaviours that fail to generalize to new or slightly modified environments. To address this, we introduce the Causal Object-centric Model Extraction Tool (COMET), a novel algorithm designed to learn the exact interpretable causal world models (CWMs). COMET first extracts object-centric state descriptions from observations and identifies the environment's internal states related to the depicted objects' properties. Using symbolic regression, it models object-centric transitions and derives causal relationships governing object dynamics. COMET further incorporates large language models (LLMs) for semantic inference, annotating causal variables to enhance interpretability. By leveraging these capabilities, COMET constructs CWMs that align with the true causal structure of the environment, enabling agents to focus on task-relevant features. The extracted CWMs mitigate the danger of shortcuts, permitting the development of RL systems capable of better planning and decision-making across dynamic scenarios. Our results, validated in Atari environments such as Pong and Freeway, demonstrate the accuracy and robustness of COMET, highlighting its potential to bridge the gap between object-centric reasoning and causal inference in reinforcement learning.

arXiv:2504.07954공식 코드 · Papers with Code

Perception-R1: Pioneering Perception Policy with Reinforcement Learning

Inspired by the success of DeepSeek-R1, we explore the potential of rule-based reinforcement learning (RL) in MLLM post-training for perception policy learning. While promising, our initial experiments reveal that incorporating a thinking process through RL does not consistently lead to performance gains across all visual perception tasks. This leads us to delve into the essential role of RL in the context of visual perception. In this work, we return to the fundamentals and explore the effects of RL on different perception tasks. We observe that the perceptual complexity is a major factor in determining the effectiveness of RL. We also observe that reward design plays a crucial role in further approching the upper limit of model perception. To leverage these findings, we propose Perception-R1, a scalable RL framework using GRPO during MLLM post-training. With a standard Qwen2.5-VL-3B-Instruct, Perception-R1 achieves +4.2% on RefCOCO+, +17.9% on PixMo-Count, +4.2% on PageOCR, and notably, 31.9% AP on COCO2017 val for the first time, establishing a strong baseline for perception policy learning.

arXiv:2504.07722

A Framework of decision-relevant observability: Reinforcement Learning converges under relative ignorability

From clinical dosing algorithms to autonomous robots, sequential decision-making systems routinely operate with missing or incomplete data. Classical reinforcement learning theory, which is commonly used to solve sequential decision problems, assumes Markovian observability, which may not hold under partial observability. Causal inference paradigms formalise ignorability of missingness. We show these views can be unified and generalized in order to guarantee Q-learning convergence even when the Markov property fails. To do so, we introduce the concept of \emph{relative ignorability}. Relative ignorability is a graphical-causal criterion which refines the requirements for accurate decision-making based on incomplete data. Theoretical results and simulations both reveal that non-markovian stochastic processes whose missingness is relatively ignorable with respect to causal estimands can still be optimized using standard Reinforcement Learning algorithms. These results expand the theoretical foundations of safe, data-efficient AI to real-world environments where complete information is unattainable.

arXiv:2504.07615공식 코드 · Papers with Code

VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model

Recently DeepSeek R1 has shown that reinforcement learning (RL) can substantially improve the reasoning capabilities of Large Language Models (LLMs) through a simple yet effective design. The core of R1 lies in its rule-based reward formulation, which leverages tasks with deterministic ground-truth answers to enable precise and stable reward computation. In the visual domain, we similarly observe that a wide range of visual understanding tasks are inherently equipped with well-defined ground-truth annotations. This property makes them naturally compatible with rule-based reward mechanisms. Motivated by this observation, we investigate the extension of R1-style reinforcement learning to Vision-Language Models (VLMs), aiming to enhance their visual reasoning capabilities. To this end, we develop VLM-R1, a dedicated framework designed to harness RL for improving VLMs' performance on general vision-language tasks. Using this framework, we further explore the feasibility of applying RL to visual domain. Experimental results indicate that the RL-based model not only delivers competitive performance on visual understanding tasks but also surpasses Supervised Fine-Tuning (SFT) in generalization ability. Furthermore, we conduct comprehensive ablation studies that uncover a series of noteworthy insights, including the presence of reward hacking in object detection, the emergence of the "OD aha moment", the impact of training data quality, and the scaling behavior of RL across different model sizes. Through these analyses, we aim to deepen the understanding of how reinforcement learning enhances the capabilities of vision-language models, and we hope our findings and open-source contributions will support continued progress in the vision-language RL community. Our code and model are available at https://github.com/om-ai-lab/VLM-R1

arXiv:2504.06983공식 코드 · Papers with Code

Free Random Projection for In-Context Reinforcement Learning

Hierarchical inductive biases are hypothesized to promote generalizable policies in reinforcement learning, as demonstrated by explicit hyperbolic latent representations and architectures. Therefore, a more flexible approach is to have these biases emerge naturally from the algorithm. We introduce Free Random Projection, an input mapping grounded in free probability theory that constructs random orthogonal matrices where hierarchical structure arises inherently. The free random projection integrates seamlessly into existing in-context reinforcement learning frameworks by encoding hierarchical organization within the input space without requiring explicit architectural modifications. Empirical results on multi-environment benchmarks show that free random projection consistently outperforms the standard random projection, leading to improvements in generalization. Furthermore, analyses within linearly solvable Markov decision processes and investigations of the spectrum of kernel random matrices reveal the theoretical underpinnings of free random projection's enhanced performance, highlighting its capacity for effective adaptation in hierarchically structured state spaces.

arXiv:2504.04469공식 코드 · Papers with Code

AI2STOW: End-to-End Deep Reinforcement Learning to Construct Master Stowage Plans under Demand Uncertainty

The worldwide economy and environmental sustainability depend on eff icient and reliable supply chains, in which container shipping plays a crucial role as an environmentally friendly mode of transport. Liner shipping companies seek to improve operational efficiency by solving the stowage planning problem. Due to many complex combinatorial aspects, stowage planning is challenging and often decomposed into two NP-hard subproblems: master and slot planning. This article proposes AI2STOW, an end-to-end deep reinforcement learning model with feasibility projection and an action mask to create master plans under demand uncertainty with global objectives and constraints, including paired block stowage patterms. Our experimental results demonstrate that AI2STOW outperforms baseline methods from reinforcement learning and stochastic programming in objective performance and computational efficiency, based on simulated instances reflecting the scale of realistic vessels and operational planning horizons.

arXiv:2504.06125공식 코드 · Papers with Code

Robo-taxi Fleet Coordination at Scale via Reinforcement Learning

Fleets of robo-taxis offering on-demand transportation services, commonly known as Autonomous Mobility-on-Demand (AMoD) systems, hold significant promise for societal benefits, such as reducing pollution, energy consumption, and urban congestion. However, orchestrating these systems at scale remains a critical challenge, with existing coordination algorithms often failing to exploit the systems' full potential. This work introduces a novel decision-making framework that unites mathematical modeling with data-driven techniques. In particular, we present the AMoD coordination problem through the lens of reinforcement learning and propose a graph network-based framework that exploits the main strengths of graph representation learning, reinforcement learning, and classical operations research tools. Extensive evaluations across diverse simulation fidelities and scenarios demonstrate the flexibility of our approach, achieving superior system performance, computational efficiency, and generalizability compared to prior methods. Finally, motivated by the need to democratize research efforts in this area, we release publicly available benchmarks, datasets, and simulators for network-level coordination alongside an open-source codebase designed to provide accessible simulation platforms and establish a standardized validation process for comparing methodologies. Code available at: https://github.com/StanfordASL/RL4AMOD

arXiv:2504.05978

Smart Exploration in Reinforcement Learning using Bounded Uncertainty Models

Reinforcement learning (RL) is a powerful tool for decision-making in uncertain environments, but it often requires large amounts of data to learn an optimal policy. We propose using prior model knowledge to guide the exploration process to speed up this learning process. This model knowledge comes in the form of a model set to which the true transition kernel and reward function belong. We optimize over this model set to obtain upper and lower bounds on the Q-function, which are then used to guide the exploration of the agent. We provide theoretical guarantees on the convergence of the Q-function to the optimal Q-function under the proposed class of exploring policies. Furthermore, we also introduce a data-driven regularized version of the model set optimization problem that ensures the convergence of the class of exploring policies to the optimal policy. Lastly, we show that when the model set has a specific structure, namely the bounded-parameter MDP (BMDP) framework, the regularized model set optimization problem becomes convex and simple to implement. In this setting, we also show that we obtain finite-time convergence to the optimal policy under additional assumptions. We demonstrate the effectiveness of the proposed exploration strategy in a simulation study. The results indicate that the proposed method can significantly speed up the learning process in reinforcement learning.

arXiv:2504.04765

Multi-Agent Deep Reinforcement Learning for Multiple Anesthetics Collaborative Control

Automated control of personalized multiple anesthetics in clinical Total Intravenous Anesthesia (TIVA) is crucial yet challenging. Current systems, including target-controlled infusion (TCI) and closed-loop systems, either rely on relatively static pharmacokinetic/pharmacodynamic (PK/PD) models or focus on single anesthetic control, limiting personalization and collaborative control. To address these issues, we propose a novel framework, Value Decomposition Multi-Agent Deep Reinforcement Learning (VD-MADRL). VD-MADRL optimizes the collaboration between two anesthetics propofol (Agent I) and remifentanil (Agent II). And It uses a Markov Game (MG) to identify optimal actions among heterogeneous agents. We employ various value function decomposition methods to resolve the credit allocation problem and enhance collaborative control. We also introduce a multivariate environment model based on random forest (RF) for anesthesia state simulation. Additionally, a data resampling and alignment technique ensures synchronized trajectory data. Our experiments on general and thoracic surgery datasets show that VD-MADRL performs better than human experience. It improves dose precision and keeps anesthesia states stable, providing great clinical value.

arXiv:2504.07095공식 코드 · Papers with Code

Neural Motion Simulator: Pushing the Limit of World Models in Reinforcement Learning

An embodied system must not only model the patterns of the external world but also understand its own motion dynamics. A motion dynamic model is essential for efficient skill acquisition and effective planning. In this work, we introduce the neural motion simulator (MoSim), a world model that predicts the future physical state of an embodied system based on current observations and actions. MoSim achieves state-of-the-art performance in physical state prediction and provides competitive performance across a range of downstream tasks. This works shows that when a world model is accurate enough and performs precise long-horizon predictions, it can facilitate efficient skill acquisition in imagined worlds and even enable zero-shot reinforcement learning. Furthermore, MoSim can transform any model-free reinforcement learning (RL) algorithm into a model-based approach, effectively decoupling physical environment modeling from RL algorithm development. This separation allows for independent advancements in RL algorithms and world modeling, significantly improving sample efficiency and enhancing generalization capabilities. Our findings highlight that world models for motion dynamics is a promising direction for developing more versatile and capable embodied systems.

arXiv:2504.05045

Attention-Augmented Inverse Reinforcement Learning with Graph Convolutions for Multi-Agent Task Allocation

This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible. Multi-agent task allocation (MATA) plays a vital role in cooperative multi-agent systems, with significant implications for applications such as logistics, search and rescue, and robotic coordination. Although traditional deep reinforcement learning (DRL) methods have been shown to be promising, their effectiveness is hindered by a reliance on manually designed reward functions and inefficiencies in dynamic environments. In this paper, an inverse reinforcement learning (IRL)-based framework is proposed, in which multi-head self-attention (MHSA) and graph attention mechanisms are incorporated to enhance reward function learning and task execution efficiency. Expert demonstrations are utilized to infer optimal reward densities, allowing dependence on handcrafted designs to be reduced and adaptability to be improved. Extensive experiments validate the superiority of the proposed method over widely used multi-agent reinforcement learning (MARL) algorithms in terms of both cumulative rewards and task execution efficiency.

arXiv:2504.04691공식 코드 · Papers with Code

Large-Scale Mixed-Traffic and Intersection Control using Multi-agent Reinforcement Learning

Traffic congestion remains a significant challenge in modern urban networks. Autonomous driving technologies have emerged as a potential solution. Among traffic control methods, reinforcement learning has shown superior performance over traffic signals in various scenarios. However, prior research has largely focused on small-scale networks or isolated intersections, leaving large-scale mixed traffic control largely unexplored. This study presents the first attempt to use decentralized multi-agent reinforcement learning for large-scale mixed traffic control in which some intersections are managed by traffic signals and others by robot vehicles. Evaluating a real-world network in Colorado Springs, CO, USA with 14 intersections, we measure traffic efficiency via average waiting time of vehicles at intersections and the number of vehicles reaching their destinations within a time window (i.e., throughput). At 80% RV penetration rate, our method reduces waiting time from 6.17 s to 5.09 s and increases throughput from 454 vehicles per 500 seconds to 493 vehicles per 500 seconds, outperforming the baseline of fully signalized intersections. These findings suggest that integrating reinforcement learning-based control large-scale traffic can improve overall efficiency and may inform future urban planning strategies.

arXiv:2504.04675

HypRL: Reinforcement Learning of Control Policies for Hyperproperties

Reward shaping in multi-agent reinforcement learning (MARL) for complex tasks remains a significant challenge. Existing approaches often fail to find optimal solutions or cannot efficiently handle such tasks. We propose HYPRL, a specification-guided reinforcement learning framework that learns control policies w.r.t. hyperproperties expressed in HyperLTL. Hyperproperties constitute a powerful formalism for specifying objectives and constraints over sets of execution traces across agents. To learn policies that maximize the satisfaction of a HyperLTL formula $\phi$, we apply Skolemization to manage quantifier alternations and define quantitative robustness functions to shape rewards over execution traces of a Markov decision process with unknown transitions. A suitable RL algorithm is then used to learn policies that collectively maximize the expected reward and, consequently, increase the probability of satisfying $\phi$. We evaluate HYPRL on a diverse set of benchmarks, including safety-aware planning, Deep Sea Treasure, and the Post Correspondence Problem. We also compare with specification-driven baselines to demonstrate the effectiveness and efficiency of HYPRL.

arXiv:2504.04160

OrbitZoo: Multi-Agent Reinforcement Learning Environment for Orbital Dynamics

The increasing number of satellites and orbital debris has made space congestion a critical issue, threatening satellite safety and sustainability. Challenges such as collision avoidance, station-keeping, and orbital maneuvering require advanced techniques to handle dynamic uncertainties and multi-agent interactions. Reinforcement learning (RL) has shown promise in this domain, enabling adaptive, autonomous policies for space operations; however, many existing RL frameworks rely on custom-built environments developed from scratch, which often use simplified models and require significant time to implement and validate the orbital dynamics, limiting their ability to fully capture real-world complexities. To address this, we introduce OrbitZoo, a versatile multi-agent RL environment built on a high-fidelity industry standard library, that enables realistic data generation, supports scenarios like collision avoidance and cooperative maneuvers, and ensures robust and accurate orbital dynamics. The environment is validated against a real satellite constellation, Starlink, achieving a Mean Absolute Percentage Error (MAPE) of 0.16% compared to real-world data. This validation ensures reliability for generating high-fidelity simulations and enabling autonomous and independent satellite operations.

arXiv:2504.03994

Improving Mixed-Criticality Scheduling with Reinforcement Learning

This paper introduces a novel reinforcement learning (RL) approach to scheduling mixed-criticality (MC) systems on processors with varying speeds. Building upon the foundation laid by [1], we extend their work to address the non-preemptive scheduling problem, which is known to be NP-hard. By modeling this scheduling challenge as a Markov Decision Process (MDP), we develop an RL agent capable of generating near-optimal schedules for real-time MC systems. Our RL-based scheduler prioritizes high-critical tasks while maintaining overall system performance. Through extensive experiments, we demonstrate the scalability and effectiveness of our approach. The RL scheduler significantly improves task completion rates, achieving around 80% overall and 85% for high-criticality tasks across 100,000 instances of synthetic data and real data under varying system conditions. Moreover, under stable conditions without degradation, the scheduler achieves 94% overall task completion and 93% for high-criticality tasks. These results highlight the potential of RL-based schedulers in real-time and safety-critical applications, offering substantial improvements in handling complex and dynamic scheduling scenarios.

arXiv:2504.03804

Offline and Distributional Reinforcement Learning for Wireless Communications

The rapid growth of heterogeneous and massive wireless connectivity in 6G networks demands intelligent solutions to ensure scalability, reliability, privacy, ultra-low latency, and effective control. Although artificial intelligence (AI) and machine learning (ML) have demonstrated their potential in this domain, traditional online reinforcement learning (RL) and deep RL methods face limitations in real-time wireless networks. For instance, these methods rely on online interaction with the environment, which might be unfeasible, costly, or unsafe. In addition, they cannot handle the inherent uncertainties in real-time wireless applications. We focus on offline and distributional RL, two advanced RL techniques that can overcome these challenges by training on static datasets and accounting for network uncertainties. We introduce a novel framework that combines offline and distributional RL for wireless communication applications. Through case studies on unmanned aerial vehicle (UAV) trajectory optimization and radio resource management (RRM), we demonstrate that our proposed Conservative Quantile Regression (CQR) algorithm outperforms conventional RL approaches regarding convergence speed and risk management. Finally, we discuss open challenges and potential future directions for applying these techniques in 6G networks, paving the way for safer and more efficient real-time wireless systems.

arXiv:2504.05521공식 코드 · Papers with Code

Deep Reinforcement Learning Algorithms for Option Hedging

Dynamic hedging is a financial strategy that consists in periodically transacting one or multiple financial assets to offset the risk associated with a correlated liability. Deep Reinforcement Learning (DRL) algorithms have been used to find optimal solutions to dynamic hedging problems by framing them as sequential decision-making problems. However, most previous work assesses the performance of only one or two DRL algorithms, making an objective comparison across algorithms difficult. In this paper, we compare the performance of eight DRL algorithms in the context of dynamic hedging; Monte Carlo Policy Gradient (MCPG), Proximal Policy Optimization (PPO), along with four variants of Deep Q-Learning (DQL) and two variants of Deep Deterministic Policy Gradient (DDPG). Two of these variants represent a novel application to the task of dynamic hedging. In our experiments, we use the Black-Scholes delta hedge as a baseline and simulate the dataset using a GJR-GARCH(1,1) model. Results show that MCPG, followed by PPO, obtain the best performance in terms of the root semi-quadratic penalty. Moreover, MCPG is the only algorithm to outperform the Black-Scholes delta hedge baseline with the allotted computational budget, possibly due to the sparsity of rewards in our environment.

arXiv:2504.03713

RLDBF: Enhancing LLMs Via Reinforcement Learning With DataBase FeedBack

While current large language models (LLMs) demonstrate remarkable linguistic capabilities through training on massive unstructured text corpora, they remain inadequate in leveraging structured scientific data (e.g., chemical molecular properties in databases) that encapsulate centuries of accumulated scientific expertise. These structured datasets hold strategic significance for advancing AI for Science yet current approaches merely treat them as auxiliary supplements to unstructured text. This study pioneers a systematic investigation into enhancing LLMs with structured scientific data, using chemical molecular science as a testbed. We investigate the impact of incorporating molecular property data on LLM across distinct training phases, including continual pre-training, supervised fine-tuning, and reinforcement learning. Notably, to address the inherent limitation of numerical insensitivity in large models, we propose an innovative methodology termed "Reinforcement Learning with Database Feedback" (RLDBF). Experimental evaluations demonstrate the efficacy of the proposed approach, with the model exhibiting remarkable generalization capabilities on previously unseen data and other chemical tasks. The results substantiate the potential of our method in advancing the field of structured scientific data processing within LLMs.

arXiv:2504.05181공식 코드 · Papers with Code

Lightweight and Direct Document Relevance Optimization for Generative Information Retrieval

Generative information retrieval (GenIR) is a promising neural retrieval paradigm that formulates document retrieval as a document identifier (docid) generation task, allowing for end-to-end optimization toward a unified global retrieval objective. However, existing GenIR models suffer from token-level misalignment, where models trained to predict the next token often fail to capture document-level relevance effectively. While reinforcement learning-based methods, such as reinforcement learning from relevance feedback (RLRF), aim to address this misalignment through reward modeling, they introduce significant complexity, requiring the optimization of an auxiliary reward function followed by reinforcement fine-tuning, which is computationally expensive and often unstable. To address these challenges, we propose direct document relevance optimization (DDRO), which aligns token-level docid generation with document-level relevance estimation through direct optimization via pairwise ranking, eliminating the need for explicit reward modeling and reinforcement learning. Experimental results on benchmark datasets, including MS MARCO document and Natural Questions, show that DDRO outperforms reinforcement learning-based methods, achieving a 7.4% improvement in MRR@10 for MS MARCO and a 19.9% improvement for Natural Questions. These findings highlight DDRO's potential to enhance retrieval effectiveness with a simplified optimization approach. By framing alignment as a direct optimization problem, DDRO simplifies the ranking optimization pipeline of GenIR models while offering a viable alternative to reinforcement learning-based methods.