English
Related papers

Related papers: Beyond Exact Gradients: Convergence of Stochastic …

200 papers

Maximum entropy deep reinforcement learning (RL) methods have been demonstrated on a range of challenging continuous tasks. However, existing methods either suffer from severe instability when training on large off-policy data or cannot…

Machine Learning · Computer Science 2019-09-10 Wenjie Shi , Shiji Song , Cheng Wu

We provide global convergence rates for vanilla and entropy-regularized federated softmax stochastic policy gradient (FedPG) with local training. We show that FedPG converges to a near-optimal policy in terms of the average agent value,…

Machine Learning · Computer Science 2026-04-02 Safwan Labbi , Paul Mangold , Daniil Tiapkin , Eric Moulines

Risk-sensitive reinforcement learning (RL) has become a popular tool for controlling the risk of uncertain outcomes and ensuring reliable performance in highly stochastic sequential decision-making problems. While it has been shown that…

Machine Learning · Computer Science 2026-01-21 Xian Yu , Lei Ying

Reinforcement learning (RL) has become a key approach for enhancing reasoning in large language models (LLMs), yet scalable training is often hindered by the rapid collapse of policy entropy, which leads to premature convergence and…

Machine Learning · Computer Science 2026-04-14 Ming Lei , Christophe Baehr

We consider stochastic gradient methods under the interpolation regime where a perfect fit can be obtained (minimum loss at each observation). While previous work highlighted the implicit regularization of such algorithms, we consider an…

Optimization and Control · Mathematics 2020-04-01 Anant Raj , Francis Bach

Policy gradient (PG) methods are a widely used reinforcement learning methodology in many applications such as video games, autonomous driving, and robotics. In spite of its empirical success, a rigorous understanding of the global…

Optimization and Control · Mathematics 2020-06-30 Kaiqing Zhang , Alec Koppel , Hao Zhu , Tamer Başar

Policy gradient (PG) methods have played an essential role in the empirical successes of reinforcement learning. In order to handle large state-action spaces, PG methods are typically used with function approximation. In this setting, the…

Since its introduction a decade ago, \emph{relative entropy policy search} (REPS) has demonstrated successful policy learning on a number of simulated and real-world robotic domains, not to mention providing algorithmic components used by…

Machine Learning · Computer Science 2021-03-18 Aldo Pacchiano , Jonathan Lee , Peter Bartlett , Ofir Nachum

We consider (stochastic) softmax policy gradient (PG) methods for bandits and tabular Markov decision processes (MDPs). While the PG objective is non-concave, recent research has used the objective's smoothness and gradient domination…

Machine Learning · Computer Science 2024-10-01 Michael Lu , Matin Aghaei , Anant Raj , Sharan Vaswani

Direct policy search serves as one of the workhorses in modern reinforcement learning (RL), and its applications in continuous control tasks have recently attracted increasing attention. In this work, we investigate the convergence theory…

Optimization and Control · Mathematics 2022-01-03 Kaiqing Zhang , Xiangyuan Zhang , Bin Hu , Tamer Başar

Policy gradient (PG) gives rise to a rich class of reinforcement learning (RL) methods. Recently, there has been an emerging trend to accelerate the existing PG methods such as REINFORCE by the \emph{variance reduction} techniques. However,…

Machine Learning · Computer Science 2021-05-31 Junyu Zhang , Chengzhuo Ni , Zheng Yu , Csaba Szepesvari , Mengdi Wang

In this effort, we consider the impact of regularization on the diversity of actions taken by policies generated from reinforcement learning agents trained using a policy gradient. Policy gradient agents are prone to entropy collapse, which…

Machine Learning · Computer Science 2023-10-10 Andrew Starnes , Anton Dereventsov , Clayton Webster

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

This paper aims to establish an entropy-regularized value-based reinforcement learning method that can ensure the monotonic improvement of policies at each policy update. Unlike previously proposed lower-bounds on policy improvement in…

Machine Learning · Computer Science 2020-08-26 Lingwei Zhu , Takamitsu Matsubara

Entropic regularization of policies in Reinforcement Learning (RL) is a commonly used heuristic to ensure that the learned policy explores the state-space sufficiently before overfitting to a local optimal policy. The primary motivation for…

Machine Learning · Computer Science 2021-01-19 Hisham Husain , Kamil Ciosek , Ryota Tomioka

Modern policy gradient algorithms, such as TRPO and PPO, outperform vanilla policy gradient in many RL tasks. Questioning the common belief that enforcing approximate trust regions leads to steady policy improvement in practice, we show…

Machine Learning · Computer Science 2025-05-27 Tao Wang , Ruipeng Zhang , Sicun Gao

We introduce the first differentiable approximation of range-partition entropy, a complexity measure from computational geometry that directly bounds algorithmic runtime. Unlike architectural modifications, our method is a complementary…

Machine Learning · Computer Science 2025-11-20 Ibne Farabi Shihab , Sanjeda Akter , Anuj Sharma

We prove that, for finite-arm bandits with linear function approximation, the global convergence of policy gradient (PG) methods depends on inter-related properties between the policy update and the representation. textcolor{blue}{First},…

Machine Learning · Computer Science 2025-04-04 Jincheng Mei , Bo Dai , Alekh Agarwal , Mohammad Ghavamzadeh , Csaba Szepesvari , Dale Schuurmans

We propose a general framework for entropy-regularized average-reward reinforcement learning in Markov decision processes (MDPs). Our approach is based on extending the linear-programming formulation of policy optimization in MDPs to…

Machine Learning · Computer Science 2017-05-23 Gergely Neu , Anders Jonsson , Vicenç Gómez

We adapt recent tools developed for the analysis of Stochastic Gradient Descent (SGD) in non-convex optimization to obtain convergence and sample complexity guarantees for the vanilla policy gradient (PG). Our only assumptions are that the…

Machine Learning · Computer Science 2022-11-21 Rui Yuan , Robert M. Gower , Alessandro Lazaric