Finite-Time Bounds for Average-Reward Fitted Q-Iteration

TL;DR

This paper introduces Anchored Fitted Q-Iteration, a novel algorithm with finite-time guarantees for average-reward offline reinforcement learning in weakly communicating MDPs, addressing a less-studied setting.

Contribution

It provides the first sample complexity analysis for average-reward offline RL with function approximation under mild assumptions, using an innovative anchor mechanism.

Findings

First finite-time bounds for average-reward offline RL

Effective in both IID and single-trajectory data settings

Introduces anchor mechanism for stability and analysis

Abstract

Although there is an extensive body of work characterizing the sample complexity of discounted-return offline RL with function approximations, prior work on the average-reward setting has received significantly less attention, and existing approaches rely on restrictive assumptions, such as ergodicity or linearity of the MDP. In this work, we establish the first sample complexity results for average-reward offline RL with function approximation for weakly communicating MDPs, a much milder assumption. To this end, we introduce Anchored Fitted Q-Iteration, which combines the standard Fitted Q-Iteration with an anchor mechanism. We show that the anchor, which can be interpreted as a form of weight decay, is crucial for enabling finite-time analysis in the average-reward setting. We also extend our finite-time analysis to the setup where the dataset is generated from a single-trajectory rather than IID transitions, again leveraging the anchor mechanism.