paper-with-me

홈 › Papers

Uniform-PAC Bounds for Reinforcement Learning with Linear Function Approximation

2021-06-22 · NeurIPS 2021 12 · Jiafan He, Dongruo Zhou, Quanquan Gu

We study reinforcement learning (RL) with linear function approximation. Existing algorithms for this problem only have high-probability regret and/or Probably Approximately Correct (PAC) sample complexity guarantees, which cannot guarantee the convergence to the optimal policy. In this paper, in order to overcome the limitation of existing algorithms, we propose a new algorithm called FLUTE, which enjoys uniform-PAC convergence to the optimal policy with high probability. The uniform-PAC guarantee is the strongest possible guarantee for reinforcement learning in the literature, which can directly imply both PAC and high probability regret bounds, making our algorithm superior to all existing algorithms with linear function approximation. At the core of our algorithm is a novel minimax value function estimator and a multi-level partition scheme to select the training samples from historical observations. Both of these techniques are new and of independent interest.

📄 PDF Abstract BibTeX arXiv:2106.11612

Code (0)

등록된 구현이 없습니다.

Tasks

reinforcement-learningReinforcement LearningReinforcement Learning (RL)

Similar Papers 제목 키워드 기반

On the Stability of Random Matrix Product with Markovian Noise: Application to Linear Stochastic Approximation and TD Learning

2021-01-30 · Alain Durmus, Eric Moulines, Alexey Naumov, Sergey Samsonov 외

This paper studies the exponential stability of random matrix products driven by a general (possibly unbounded) state space Markov chain. It is a cornerstone in the analysis of stochastic algorithms in machine learning (…

reinforcement-learningReinforcement LearningReinforcement Learning (RL)

Uniform-PAC Guarantees for Model-Based RL with Bounded Eluder Dimension

2023-05-15 · Yue Wu, Jiafan He, Quanquan Gu

Recently, there has been remarkable progress in reinforcement learning (RL) with general function approximation. However, all these works only provide regret or sample complexity guarantees. It is still an open question …

Open-Ended Question AnsweringReinforcement Learning (RL)

On Sharpness of Error Bounds for Multivariate Neural Network Approximation

2020-04-05 · Steffen Goebbels

Single hidden layer feedforward neural networks can represent multivariate functions that are sums of ridge functions. These ridge functions are defined via an activation function and customizable weights. The paper deal…

Math

Provably Efficient Model-Free Constrained RL with Linear Function Approximation

2022-06-23 · Arnob Ghosh, Xingyu Zhou, Ness Shroff

We study the constrained reinforcement learning problem, in which an agent aims to maximize the expected cumulative reward subject to a constraint on the expected total value of a utility function. In contrast to existin…

Sharp Lower Bounds for Linearized ReLU^k Approximation on the Sphere

2025-10-05 · Tong Mao, Jinchao Xu arxiv

We prove a saturation theorem for linearized shallow ReLU$^k$ neural networks on the unit sphere $\mathbb S^d$. For any antipodally quasi-uniform set of centers, if the target function has smoothness $r>\tfrac{d+2k+1}{2}…