부산대학교 도서관

Reinforcement Learning allows us to acquire knowledge without any training data. However, for learning it takes time. We discuss a case in which an agent receives a large negative reward. We assume that the reverse action allows us to improve the current situation. In this work, we propose a method to perform Reverse action by using Retrospective Kalman Filter that estimates the state one step before. We show an experience by a Hunter Prey problem. And discuss the usefulness of our proposed method.