Efficient Reinforcement Learning in Deterministic Systems with Value Function Generalization

TL;DR

This paper introduces an algorithm called optimistic constraint propagation (OCP) for efficient reinforcement learning in deterministic systems, providing theoretical guarantees and computational insights.

Contribution

The paper proposes OCP, a novel algorithm that combines exploration and value function generalization with theoretical performance bounds.

Findings

OCP achieves near-optimal performance in finite episodes when the true value function is within the hypothesis class.

Theoretical guarantees extend to cases where the true value function is outside the hypothesis class, under specific conditions.

Computational results demonstrate the practical effectiveness of OCP in illustrative examples.

Abstract

We consider the problem of reinforcement learning over episodes of a finite-horizon deterministic system and as a solution propose optimistic constraint propagation (OCP), an algorithm designed to synthesize efficient exploration and value function generalization. We establish that when the true value function lies within a given hypothesis class, OCP selects optimal actions over all but at most K episodes, where K is the eluder dimension of the given hypothesis class. We establish further efficiency and asymptotic performance guarantees that apply even if the true value function does not lie in the given hypothesis class, for the special case where the hypothesis class is the span of pre-specified indicator functions over disjoint sets. We also discuss the computational complexity of OCP and present computational results involving two illustrative examples.