English
Related papers

Related papers: Policy Mirror Descent with Lookahead

200 papers

We study the sequential decision making problem of maximizing the expected total reward while satisfying a constraint on the expected total utility. We employ the natural policy gradient method to solve the discounted infinite-horizon…

Optimization and Control · Mathematics 2025-10-16 Dongsheng Ding , Kaiqing Zhang , Jiali Duan , Tamer Başar , Mihailo R. Jovanović

Partially observable Markov decision processes (POMDPs) model specific environments in sequential decision-making under uncertainty. Critically, optimal policies for POMDPs may not be robust against perturbations in the environment.…

Artificial Intelligence · Computer Science 2025-08-21 Maris F. L. Galesloot , Roman Andriushchenko , Milan Češka , Sebastian Junges , Nils Jansen

To facilitate efficient learning, policy gradient approaches to deep reinforcement learning (RL) are typically paired with variance reduction measures and strategies for making large but safe policy changes based on a batch of experiences.…

Machine Learning · Computer Science 2023-11-13 Jared Markowitz , Edward W. Staley

This note re-visits the rolling-horizon control approach to the problem of a Markov decision process (MDP) with infinite-horizon discounted expected reward criterion. Distinguished from the classical value-iteration approach, we develop an…

Optimization and Control · Mathematics 2022-06-07 Hyeong Soo Chang

This work revisits standard policy gradient methods used on restricted policy classes, which are known to get stuck in suboptimal critical points. We identify an important cause for this phenomenon to be that the policy gradient is itself…

Machine Learning · Computer Science 2026-05-12 Alex DeWeese , Guannan Qu

Policy Dual Averaging (PDA) offers a principled Policy Mirror Descent (PMD) framework that more naturally admits value function approximation than standard PMD, enabling the use of approximate advantage (or Q-) functions while retaining…

Machine Learning · Computer Science 2026-03-12 Ji Gao , Caleb Ju , Guanghui Lan , Zhaohui Tong

The classical policy gradient method is the theoretical and conceptual foundation of modern policy-based reinforcement learning (RL) algorithms. Most rigorous analyses of such methods, particularly those establishing convergence guarantees,…

Machine Learning · Computer Science 2026-02-11 Jongmin Lee , Ernest K. Ryu

Pruning is an effective method for compressing Large Language Models, but finding an optimal, non-uniform layer-wise sparsity allocation remains a key challenge. While heuristic methods are fast but yield suboptimal performance, more…

Machine Learning · Computer Science 2025-11-25 Xin Yuan , Siqi Li , Jiateng Wei , Chengrui Zhu , Yanming Wu , Qingpeng Li , Jiajun Lv , Xiaoke Lan , Jun Chen , Yong Liu

Recent theory suggests that reward-model-first methods can be more sample-efficient than direct policy fitting when the reward function is statistically simpler than the induced policy. We propose DDO-RM, a finite-candidate…

Machine Learning · Statistics 2026-05-01 Tiantian Zhang , Jierui Zuo , Michael Chen , Wenping Wang

Markov Decision Processes (MDPs) are a formal framework for modeling and solving sequential decision-making problems. In finite-time horizons such problems are relevant for instance for optimal stopping or specific supply chain problems,…

Optimization and Control · Mathematics 2024-05-07 Sara Klein , Simon Weissmann , Leif Döring

We build on a recently introduced geometric interpretation of Markov Decision Processes (MDPs) to analyze classical MDP-solving algorithms: Value Iteration (VI) and Policy Iteration (PI). First, we develop a geometry-based analytical…

Machine Learning · Computer Science 2025-03-07 Arsenii Mustafin , Aleksei Pakharev , Alex Olshevsky , Ioannis Ch. Paschalidis

We introduce a method for policy improvement that interpolates between the greedy approach of value-based reinforcement learning (RL) and the full planning approach typical of model-based RL. The new method builds on the concept of a…

Machine Learning · Statistics 2022-06-20 Shantanu Thakoor , Mark Rowland , Diana Borsa , Will Dabney , Rémi Munos , André Barreto

We consider off-policy temporal-difference (TD) learning methods for policy evaluation in Markov decision processes with finite spaces and discounted reward criteria, and we present a collection of convergence results for several…

Machine Learning · Computer Science 2018-03-30 Huizhen Yu

Solving goal-oriented tasks is an important but challenging problem in reinforcement learning (RL). For such tasks, the rewards are often sparse, making it difficult to learn a policy effectively. To tackle this difficulty, we propose a new…

Machine Learning · Computer Science 2019-11-04 Hao Sun , Zhizhong Li , Xiaotong Liu , Dahua Lin , Bolei Zhou

Approaches to policy optimization have been motivated from diverse principles, based on how the parametric model is interpreted (e.g. value versus policy representation) or how the learning objective is formulated, yet they share a common…

Machine Learning · Computer Science 2022-06-20 Ramki Gummadi , Saurabh Kumar , Junfeng Wen , Dale Schuurmans

This paper proposes a payoff perturbation technique for the Mirror Descent (MD) algorithm in games where the gradient of the payoff functions is monotone in the strategy profile space, potentially containing additive noise. The optimistic…

Computer Science and Game Theory · Computer Science 2024-06-25 Kenshi Abe , Kaito Ariu , Mitsuki Sakamoto , Atsushi Iwasaki

Learning Markov decision processes (MDPs) in the presence of the adversary is a challenging problem in reinforcement learning (RL). In this paper, we study RL in episodic MDPs with adversarial reward and full information feedback, where the…

Machine Learning · Computer Science 2022-04-21 Jiafan He , Dongruo Zhou , Quanquan Gu

We study a primal-dual (PD) reinforcement learning (RL) algorithm for online constrained Markov decision processes (CMDPs). Despite its widespread practical use, the existing theoretical literature on PD-RL algorithms for this problem only…

Modern deep reinforcement learning (RL) algorithms are motivated by either the generalised policy iteration (GPI) or trust-region learning (TRL) frameworks. However, algorithms that strictly respect these theoretical frameworks have proven…

Machine Learning · Computer Science 2024-11-21 Jakub Grudzien Kuba , Christian Schroeder de Witt , Jakob Foerster

The softmax policy gradient (PG) method, which performs gradient ascent under softmax policy parameterization, is arguably one of the de facto implementations of policy optimization in modern reinforcement learning. For $\gamma$-discounted…

Machine Learning · Computer Science 2022-12-19 Gen Li , Yuting Wei , Yuejie Chi , Yuxin Chen