Title: A Study for Improvement for Reinforcement Learning based on Knowledge Sharing Method — Adaptability to a situation of intermingled of complete and incomplete perception under an maze—

Issue Number: Vol. 9, No. 2
Year of Publication: Jun - 2019
Page Numbers: 60-67
Authors: Masashi SUGIMOTO, Hiroya YASHIRO, Kazuma NISHIMURA, Shinji TSUZUKI, Kentarou KURASHIGE
Journal Name: International Journal of New Computer Architectures and their Applications (IJNCAA)
- Hong Kong

Abstract:


This study aims to improve maze-solving technique, efficiency by which to the reinforcement learning agent under the situation of using incomplete sensors. In Reinforcement Learning, this method will be supposed that agent is able to observe the environment, completely. However, there is a limit on the information of the sensors. Moreover, it is hard to learn the Reinforcement learning agent in the actual environment cause of some noise of actual environment or source device. In this study, the proposed method has been using two types agent that included complete perception and incomplete perception and exchange of information on the location to settle this problem. This study aims to improve maze-solving technique, efficiency by which to the reinforcement learning agent under the situation of using incomplete sensors. As a result, the proposed method has been confirmed that is provided suitable solution for an approach to the goal for incomplete agents.