PEGASUS: A Policy Search Method for Large MDPs and POMDPs

TL;DR

This paper introduces PEGASUS, a policy search method for large MDPs and POMDPs that transforms the problem into deterministic POMDPs, enabling efficient policy evaluation and search with polynomial sample complexity bounds.

Contribution

The paper presents a novel policy search approach that transforms (PO)MDPs into deterministic equivalents, allowing for effective value estimation and policy optimization with theoretical guarantees.

Findings

Effective policy search in large POMDPs demonstrated on complex problems.

Polynomial sample complexity bounds established for the method.

Empirical results show success on both discrete and continuous domains.

Abstract

We propose a new approach to the problem of searching a space of policies for a Markov decision process (MDP) or a partially observable Markov decision process (POMDP), given a model. Our approach is based on the following observation: Any (PO)MDP can be transformed into an "equivalent" POMDP in which all state transitions (given the current state and action) are deterministic. This reduces the general problem of policy search to one in which we need only consider POMDPs with deterministic transitions. We give a natural way of estimating the value of all policies in these transformed POMDPs. Policy search is then simply performed by searching for a policy with high estimated value. We also establish conditions under which our value estimates will be good, recovering theoretical results similar to those of Kearns, Mansour and Ng (1999), but with "sample complexity" bounds that have only a polynomial rather than exponential dependence on the horizon time. Our method applies to arbitrary POMDPs, including ones with infinite state and action spaces. We also present empirical results for our approach on a small discrete problem, and on a complex continuous state/continuous action problem involving learning to ride a bicycle.