TY - GEN
T1 - Context-aware policy reuse
AU - Li, Siyuan
AU - Gu, Fangda
AU - Zhu, Guangxiang
AU - Zhang, Chongjie
N1 - Publisher Copyright:
© 2019 International Foundation for Autonomous Agents and Multiagent Systems (www.ifaamas.org) Ail rights reserved.
PY - 2019
Y1 - 2019
N2 - Transfer learning can greatly speed up reinforcement learning for a new task by leveraging policies of relevant tasks Existing works of policy reuse either focus on selecting a single best source policy for reuse without considering contexts, or fail to guarantee learning an optimal policy for a target task To improve transfer efficiency and guarantee optimality, we develop a novel policy reuse method, called Context-Aware Policy reuSe (CAPS), that enables multi-policy reuse Our method learns when and which source policy is best for reuse, as well as when to terminate its reuse CAPS provides theoretical guarantees in convergence and optimality for both source policy selection and target task learning Empirical results on a grid-based navigation domain and the Pygame Learning Environment demonstrate that CAPS significantly outperforms other state-of-the-art policy reuse methods.
AB - Transfer learning can greatly speed up reinforcement learning for a new task by leveraging policies of relevant tasks Existing works of policy reuse either focus on selecting a single best source policy for reuse without considering contexts, or fail to guarantee learning an optimal policy for a target task To improve transfer efficiency and guarantee optimality, we develop a novel policy reuse method, called Context-Aware Policy reuSe (CAPS), that enables multi-policy reuse Our method learns when and which source policy is best for reuse, as well as when to terminate its reuse CAPS provides theoretical guarantees in convergence and optimality for both source policy selection and target task learning Empirical results on a grid-based navigation domain and the Pygame Learning Environment demonstrate that CAPS significantly outperforms other state-of-the-art policy reuse methods.
KW - Policy reuse
KW - Reinforcement learning
KW - Transfer learning
UR - https://www.scopus.com/pages/publications/85077002037
M3 - Conference contribution
AN - SCOPUS:85077002037
T3 - Proceedings of the International Joint Conference on Autonomous Agents and Multiagent Systems, AAMAS
SP - 989
EP - 997
BT - 18th International Conference on Autonomous Agents and Multiagent Systems, AAMAS 2019
PB - International Foundation for Autonomous Agents and Multiagent Systems (IFAAMAS)
T2 - 18th International Conference on Autonomous Agents and Multiagent Systems, AAMAS 2019
Y2 - 13 May 2019 through 17 May 2019
ER -