Basic Research on Transfer Learning Indicators for Reinforcement Learning

Reinforcement learning requires a lot of time for the agent to learn. Transfer learning methods can be used to shorten this learning time, but they have the disadvantage that it is not known which knowledge is effective in what kind of environment until it is learned. When users transfer knowledge, it is necessary to investigate the relationship between the transfer source and the transfer destination. This study proposes an adaptive criteria evaluation index that can determine this relationship in advance. In the simulation, we confirmed the effectiveness of the proposed method using several problem examples. (c) 2022The Author. Published by Sugisaka Masanori at ALife Robotics Corporation Ltd.This is an open access article distributed under the CC BY-NC 4.0 license