Agnostic System Identification for Model-Based Reinforcement Learning

TL;DR

This paper introduces an iterative, agnostic system identification method for model-based reinforcement learning that guarantees near-optimal policies even when the true system isn't in the considered model class, applicable to discrete and continuous domains.

Contribution

It presents a novel agnostic approach using no-regret online learning algorithms for robust policy learning without assuming the true system is in the model class.

Findings

Effective in both discrete and continuous domains.

Scalable and demonstrated on helicopter control tasks.

Provides strong guarantees even when the true system is outside the model class.

Abstract

A fundamental problem in control is to learn a model of a system from observations that is useful for controller synthesis. To provide good performance guarantees, existing methods must assume that the real system is in the class of models considered during learning. We present an iterative method with strong guarantees even in the agnostic case where the system is not in the class. In particular, we show that any no-regret online learning algorithm can be used to obtain a near-optimal policy, provided some model achieves low training error and access to a good exploration distribution. Our approach applies to both discrete and continuous domains. We demonstrate its efficacy and scalability on a challenging helicopter domain from the literature.