Efficient Deep Reinforcement Learning via Adaptive Policy Transfer

TL;DR

This paper introduces a novel Policy Transfer Framework that optimizes target policies in deep reinforcement learning by adaptively selecting and terminating source policies without explicit similarity measures, leading to faster learning.

Contribution

It presents a new framework that models multi-policy transfer as an option learning problem, enabling adaptive policy reuse without explicit task similarity computation.

Findings

Significantly accelerates RL learning process.

Outperforms state-of-the-art transfer methods in efficiency.

Achieves better final performance in various action spaces.

Abstract

Transfer Learning (TL) has shown great potential to accelerate Reinforcement Learning (RL) by leveraging prior knowledge from past learned policies of relevant tasks. Existing transfer approaches either explicitly computes the similarity between tasks or select appropriate source policies to provide guided explorations for the target task. However, how to directly optimize the target policy by alternatively utilizing knowledge from appropriate source policies without explicitly measuring the similarity is currently missing. In this paper, we propose a novel Policy Transfer Framework (PTF) to accelerate RL by taking advantage of this idea. Our framework learns when and which source policy is the best to reuse for the target policy and when to terminate it by modeling multi-policy transfer as the option learning problem. PTF can be easily combined with existing deep RL approaches. Experimental results show it significantly accelerates the learning process and surpasses state-of-the-art policy transfer methods in terms of learning efficiency and final performance in both discrete and continuous action spaces.