Transfer from Multiple MDPs

TL;DR

This paper explores transfer reinforcement learning from multiple source Markov Decision Processes (MDPs), analyzing its theoretical properties, proposing adaptive algorithms based on task similarity, and demonstrating effectiveness through experiments.

Contribution

It introduces novel algorithms that adapt transfer based on task similarity and provides a theoretical analysis of transfer from multiple MDPs.

Findings

Theoretical properties of transfer from multiple MDPs are characterized.

Adaptive algorithms improve transfer efficiency based on task similarity.

Experimental results validate the proposed methods on a continuous chain problem.

Abstract

Transfer reinforcement learning (RL) methods leverage on the experience collected on a set of source tasks to speed-up RL algorithms. A simple and effective approach is to transfer samples from source tasks and include them into the training set used to solve a given target task. In this paper, we investigate the theoretical properties of this transfer method and we introduce novel algorithms adapting the transfer process on the basis of the similarity between source and target tasks. Finally, we report illustrative experimental results in a continuous chain problem.