Reward-Free Model-Based Reinforcement Learning with Linear Function Approximation

Abstract

We study the model-based reward-free reinforcement learning with linear function approximation for episodic Markov decision processes (MDPs). In this setting, the agent works in two phases. In the exploration phase, the agent interacts with the environment and collects samples without the reward. In the planning phase, the agent is given a specific reward function and uses samples collected from the exploration phase to learn a good policy. We propose a new provably efficient algorithm, called UCRL-RFE under the Linear Mixture MDP assumption, where the transition probability kernel of the MDP can be parameterized by a linear function over certain feature mappings defined on the triplet of state, action, and next state. We show that to obtain an $\epsilon$-optimal policy for arbitrary reward function, UCRL-RFE needs to sample at most $\tilde{\mathcal{O}}(H^5{d}^2{\epsilon }^{-2})$ episodes during the exploration phase. Here, $H$ is the length of the episode, $d$ is the dimension of the feature mapping. We also propose a variant of UCRL-RFE using Bernstein-type bonus and show that it needs to sample at most $\tilde{\mathcal{O}}(H^{4}d(H+d){\epsilon }^{-2})$ to achieve an $\epsilon$-optimal policy. By constructing a special class of linear Mixture MDPs, we also prove that for any reward-free algorithm, it needs to sample at least $\tilde{\Omega }(H^{2}d{\epsilon }^{-2})$ episodes to obtain an $\epsilon$-optimal policy. Our upper bound matches the lower bound in terms of the dependence on $\epsilon$ and the dependence on $d$ if $H\ge d$.

29 pages, 1 figure, 1 table. In NeurIPS 2021

References in corpus (8)

• Model-Based Reinforcement Learning with Value-Targeted Regression
• Learning Near Optimal Policies with Low Inherent Bellman Error
• Optimism in Reinforcement Learning with Generalized Linear Function Approximation
• Fast active learning for pure exploration in reinforcement learning
• Reward-Free Exploration for Reinforcement Learning
• Nearly Minimax Optimal Reward-free Reinforcement Learning
• A Sharp Analysis of Model-based Reinforcement Learning with Self-Play
• Nearly Minimax Optimal Regret for Learning Infinite-horizon Average-reward MDPs with Linear Function Approximation

Cited by in corpus (2)

• Near-Optimal Reward-Free Exploration for Linear Mixture MDPs with Plug-in Solver
• Adaptive Multi-Goal Exploration