@inproceedings{f6d4343f522e4ae79930b18002c83638,
title = "Selective Transition Collection in Experience Replay",
abstract = "Experience replay method is often used in off-policy reinforcement learning. As the training progresses, the distribution of the collected transitions becomes more and more concentrated, and this will lead to catastrophic forgetting and a low rate of convergence. In this paper, we present selective transition collection algorithm which is a new design to address the concentrated distribution by selectively collection the transitions. We propose a method to estimate the similarity between transitions, and a probability function to reduce the chance of transitions with high similarity to the experience memory being collected. We test our method on familiar reinforcement learning tasks and the experimental results demonstrate that selective transition collection can not only speed up the learning but also prevent catastrophic forgetting effectively.",
keywords = "Experience replay, Reinforcement learning, Selective transition collection",
author = "Feng Liu and Shuling Dai and Yongjia Zhao",
note = "Publisher Copyright: {\textcopyright} 2021, Springer Nature Singapore Pte Ltd.; 5th International Conference on Cognitive Systems and Signal Processing, ICCSIP 2020 ; Conference date: 25-12-2020 Through 27-12-2020",
year = "2021",
doi = "10.1007/978-981-16-2336-3\_29",
language = "英语",
isbn = "9789811623356",
series = "Communications in Computer and Information Science",
publisher = "Springer Science and Business Media Deutschland GmbH",
pages = "317--324",
editor = "Fuchun Sun and Huaping Liu and Bin Fang",
booktitle = "Cognitive Systems and Signal Processing - 5th International Conference, ICCSIP 2020, Revised Selected Papers",
address = "德国",
}