中文
相关论文

相关论文: Finite-Sample Analysis of Off-Policy TD-Learning v…

200 篇论文

Continuous-time reinforcement learning (CTRL) provides a principled framework for sequential decision-making in environments where interactions evolve continuously over time. Despite its empirical success, the theoretical understanding of…

机器学习 · 计算机科学 2025-05-22 Runze Zhao , Yue Yu , Adams Yiyue Zhu , Chen Yang , Dongruo Zhou

Aligning large language models (LLMs) on domain-specific data remains a fundamental challenge. Supervised fine-tuning (SFT) offers a straightforward way to inject domain knowledge but often degrades the model's generality. In contrast,…

机器学习 · 计算机科学 2026-02-12 Linxuan Xia , Xiaolong Yang , Yongyuan Chen , Enyue Zhao , Deng Cai , Yasheng Wang , Boxi Wu

Many reinforcement learning algorithms, particularly those that rely on return estimates for policy improvement, can suffer from poor sample efficiency and training instability due to high-variance return estimates. In this paper we…

机器学习 · 计算机科学 2026-01-06 Alexander W. Goodall , Edwin Hamel-De le Court , Francesco Belardinelli

We consider off-policy policy evaluation with function approximation (FA) in average-reward MDPs, where the goal is to estimate both the reward rate and the differential value function. For this problem, bootstrapping is necessary and,…

机器学习 · 计算机科学 2022-10-19 Shangtong Zhang , Yi Wan , Richard S. Sutton , Shimon Whiteson

In many real-world settings, reinforcement learning systems suffer performance degradation when the environment encountered at deployment differs from that observed during training. Distributionally robust reinforcement learning (DR-RL)…

机器学习 · 计算机科学 2026-03-05 Debamita Ghosh , George K. Atia , Yue Wang

We propose a novel distributionally robust $Q$-learning algorithm for the non-tabular case accounting for continuous state spaces where the state transition of the underlying Markov decision process is subject to model uncertainty. The…

机器学习 · 计算机科学 2025-05-27 Chung I Lu , Julian Sester , Aijia Zhang

Reinforcement learning for large language models faces a fundamental trade-off between sample efficiency and asymptotic performance: strictly on-policy methods discard trajectories after a single update, while off-policy reuse introduces…

机器学习 · 计算机科学 2026-05-26 Changyu Chen , Xiting Wang , Rui Yan

Off-policy estimation for long-horizon problems is important in many real-life applications such as healthcare and robotics, where high-fidelity simulators may not be available and on-policy evaluation is expensive or impossible. Recently,…

机器学习 · 计算机科学 2020-03-26 Ali Mousavi , Lihong Li , Qiang Liu , Denny Zhou

Multi-step temporal difference (TD) learning is an important approach in reinforcement learning, as it unifies one-step TD learning with Monte Carlo methods in a way where intermediate algorithms can outperform either extreme. They address…

机器学习 · 计算机科学 2018-09-10 Kristopher De Asis , Richard S. Sutton

Gradient temporal-difference (GTD) learning algorithms are widely used for off-policy policy evaluation with function approximation. However, existing convergence analyses rely on the restrictive assumption that the so-called feature…

机器学习 · 计算机科学 2026-05-11 Hyunjun Na , Donghwan Lee

Motivated by applications in risk-sensitive reinforcement learning, we study mean-variance optimization in a discounted reward Markov Decision Process (MDP). Specifically, we analyze a Temporal Difference (TD) learning algorithm with linear…

机器学习 · 计算机科学 2025-03-13 Tejaram Sangadi , L. A. Prashanth , Krishna Jagannathan

Distributional reinforcement learning improves performance by capturing environmental stochasticity, but a comprehensive theoretical understanding of its effectiveness remains elusive. In addition, the intractable element of the infinite…

机器学习 · 计算机科学 2025-05-14 Taehyun Cho , Seungyub Han , Seokhun Ju , Dohyeong Kim , Kyungjae Lee , Jungwoo Lee

Offline reinforcement learning (RL) has received increasing attention for learning policies from previously collected data without interaction with the real environment, which is particularly important in high-stakes applications. While a…

机器学习 · 计算机科学 2026-04-03 Qing Zhu , Xian Yu

Temporal-Difference (TD) learning is a general and very useful tool for estimating the value function of a given policy, which in turn is required to find good policies. Generally speaking, TD learning updates states whenever they are…

机器学习 · 计算机科学 2021-08-24 Nishanth Anand , Doina Precup

Off-policy reinforcement learning (RL) using a fixed offline dataset of logged interactions is an important consideration in real world applications. This paper studies offline RL using the DQN replay dataset comprising the entire replay…

机器学习 · 计算机科学 2020-11-25 Rishabh Agarwal , Dale Schuurmans , Mohammad Norouzi

We propose a refinement of temporal-difference learning that enforces first-order Bellman consistency: the learned value function is trained to match not only the Bellman targets in value but also their derivatives with respect to states…

机器学习 · 计算机科学 2025-11-25 Fabian Schramm , Nicolas Perrin-Gilbert , Justin Carpentier

We study the multi-step off-policy learning approach to distributional RL. Despite the apparent similarity between value-based RL and distributional RL, our study reveals intriguing and fundamental differences between the two cases in the…

机器学习 · 计算机科学 2022-07-18 Yunhao Tang , Mark Rowland , Rémi Munos , Bernardo Ávila Pires , Will Dabney , Marc G. Bellemare

We study offline reinforcement learning (RL) which seeks to learn a good policy based on a fixed, pre-collected dataset. A fundamental challenge behind this task is the distributional shift due to the dataset lacking sufficient exploration,…

机器学习 · 计算机科学 2023-10-11 Wenzhuo Zhou

The linear programming (LP) approach has a long history in the theory of approximate dynamic programming. When it comes to computation, however, the LP approach often suffers from poor scalability. In this work, we introduce a relaxed…

系统与控制 · 电气工程与系统科学 2020-12-01 Andrea Martinelli , Matilde Gargiani , John Lygeros

We study the problem of infinite-horizon average-reward reinforcement learning with linear Markov decision processes (MDPs). The associated Bellman operator of the problem not being a contraction makes the algorithm design challenging.…

机器学习 · 统计学 2025-03-12 Kihyuk Hong , Woojin Chae , Yufan Zhang , Dabeen Lee , Ambuj Tewari