Planning with RL and episodic-memory behavioral priors

TL;DR

This paper introduces a planning-based reinforcement learning method that leverages behavioral priors to improve exploration efficiency and learning speed, addressing limitations of existing imitation learning approaches.

Contribution

It presents a novel planning approach that effectively incorporates behavioral priors, reducing the need for extensive expert demonstrations and complex learning algorithms.

Findings

Behavioral priors enhance exploration efficiency.

The proposed method accelerates learning in RL environments.

Curated exploration policies improve sample efficiency.

Abstract

The practical application of learning agents requires sample efficient and interpretable algorithms. Learning from behavioral priors is a promising way to bootstrap agents with a better-than-random exploration policy or a safe-guard against the pitfalls of early learning. Existing solutions for imitation learning require a large number of expert demonstrations and rely on hard-to-interpret learning methods like Deep Q-learning. In this work we present a planning-based approach that can use these behavioral priors for effective exploration and learning in a reinforcement learning environment, and we demonstrate that curated exploration policies in the form of behavioral priors can help an agent learn faster.