English
Related papers

Related papers: Is the Policy Gradient a Gradient?

200 papers

Policy gradient methods are widely used in reinforcement learning. Yet, the nonconvexity of policy optimization poses significant challenges in understanding the global convergence of policy gradient methods. For a class of finite-horizon…

Optimization and Control · Mathematics 2026-03-10 Xin Chen , Yifan Hu , Minda Zhao

Neural networks trained with standard objectives exhibit behaviors characteristic of probabilistic inference: soft clustering, prototype specialization, and Bayesian uncertainty tracking. These phenomena appear across architectures -- in…

Machine Learning · Computer Science 2026-01-01 Alan Oursland

Standard policy gradients weight each sampled action by advantage alone, regardless of how likely that action was under the current policy. This creates two pathologies: within a single decision context (e.g. one image or prompt), a rare…

Machine Learning · Computer Science 2026-03-17 Ian Osband

The study of first-order optimization is sensitive to the assumptions made on the objective functions. These assumptions induce complexity classes which play a key role in worst-case analysis, including the fundamental concept of algorithm…

Optimization and Control · Mathematics 2024-05-30 Charles Guille-Escuret , Adam Ibrahim , Baptiste Goujaud , Ioannis Mitliagkas

Policy gradient based reinforcement learning algorithms coupled with neural networks have shown success in learning complex policies in the model free continuous action space control setting. However, explicitly parameterized policies are…

Machine Learning · Computer Science 2019-09-30 Oliver Richter , Roger Wattenhofer

When designing algorithms for finite-time-horizon episodic reinforcement learning problems, a common approach is to introduce a fictitious discount factor and use stationary policies for approximations. Empirically, it has been shown that…

Machine Learning · Computer Science 2021-09-15 Xin Guo , Anran Hu , Junzi Zhang

This paper proposes a distributed algorithm for a network of agents to solve an optimization problem with separable objective function and locally coupled constraints. Our strategy is based on reformulating the original constrained problem…

Optimization and Control · Mathematics 2021-03-12 Priyank Srivastava , Jorge Cortes

This paper investigates the problem of online prediction learning, where learning proceeds continuously as the agent interacts with an environment. The predictions made by the agent are contingent on a particular way of behaving,…

Machine Learning · Computer Science 2018-11-08 Sina Ghiassian , Andrew Patterson , Martha White , Richard S. Sutton , Adam White

Learning optimal behavior from existing data is one of the most important problems in Reinforcement Learning (RL). This is known as "off-policy control" in RL where an agent's objective is to compute an optimal policy based on the data…

Machine Learning · Computer Science 2022-06-16 Raghuram Bharadwaj Diddigi , Prateek Jain , Prabuchandran K. J. , Shalabh Bhatnagar

Adaptive gradient methods have achieved remarkable success in training deep neural networks on a wide variety of tasks. However, not much is known about the mathematical and statistical properties of this family of methods. This work aims…

Machine Learning · Computer Science 2021-05-18 Zhang Zhiyi , Liu Ziyin

We present the Generalized Policy Gradient (GPG) Theorem, specifically designed for Transformer-based policies. Notably, we demonstrate that both standard Policy Gradient Theorem and GRPO emerge as special cases within our GPG framework.…

Machine Learning · Computer Science 2025-12-12 Hangyu Mao , Guangting Dong , Zhicheng Dou

Option-critic learning is a general-purpose reinforcement learning (RL) framework that aims to address the issue of long term credit assignment by leveraging temporal abstractions. However, when dealing with extended timescales, discounting…

Machine Learning · Computer Science 2019-11-21 Akshay Dharmavaram , Matthew Riemer , Shalabh Bhatnagar

Reinforcement learning methods for robotics are increasingly successful due to the constant development of better policy gradient techniques. A precise (low variance) and accurate (low bias) gradient estimator is crucial to face…

Machine Learning · Computer Science 2022-03-09 Joao Carvalho , Jan Peters

Stochastic gradient descent (SGD) is a powerful optimization technique that is particularly useful in online learning scenarios. Its convergence analysis is relatively well understood under the assumption that the data samples are…

Machine Learning · Computer Science 2024-10-03 Ethan Che , Jing Dong , Xin T. Tong

Communication is fundamental to sustaining reciprocity and cooperation in strategic interactions. We identify and formulate the influence attribution problem as the central optimization difficulty inherent in such dynamics for a learning…

Machine Learning · Computer Science 2026-05-12 Yue Lin , Pascal Poupart , Shuhui Zhu , Dan Qiao , Wenhao Li , Yuan Liu , Hongyuan Zha , Baoxiang Wang

This paper proposes a scalable distributed policy gradient method and proves its convergence to near-optimal solution in multi-agent linear quadratic networked systems. The agents engage within a specified network under local communication…

Multiagent Systems · Computer Science 2024-03-06 Yuzi Yan , Yuan Shen

The increasing number of wireless devices operating in unlicensed spectrum motivates the development of intelligent adaptive approaches to spectrum access that go beyond traditional carrier sensing. We develop a novel distributed…

Networking and Internet Architecture · Computer Science 2021-11-19 Akash Doshi , Jeffrey G. Andrews

The policy gradient theorem states that the policy should only be updated in states that are visited by the current policy, which leads to insufficient planning in the off-policy states, and thus to convergence to suboptimal policies. We…

Machine Learning · Computer Science 2021-10-01 Romain Laroche , Remi Tachet

This paper considers non-smooth optimization problems where we seek to minimize the pointwise maximum of a continuously parameterized family of functions. Since the objective function is given as the solution to a maximization problem,…

Optimization and Control · Mathematics 2026-01-12 Dimitris Boskos , Jorge Cortés , Sonia Martínez

We explore reinforcement learning methods for finding the optimal policy in the linear quadratic regulator (LQR) problem. In particular, we consider the convergence of policy gradient methods in the setting of known and unknown parameters.…

Machine Learning · Computer Science 2021-06-25 Ben Hambly , Renyuan Xu , Huining Yang