par Coppens, Youri ;Shirahata, Koichi;Fukagai, Takuya;Tomita, Yasumoto;Ike, Atsushi
Référence 2017 Fifth International Symposium on Computing and Networking (CANDAR)(19-22 Nov. 2017: Aomori, Japan), 2017 Fifth International Symposium on Computing and Networking (CANDAR), IEEE, page (330-336)
Publication Publié, 2017-11-19
Publication dans des actes
Résumé : Recent state-of-the-art Deep Reinforcement Learn- ing algorithms, such as A3C and UNREAL, are designed to train on a single device with only CPU's. Using GPU acceleration for these algorithms results in low GPU utilization, which means the full performance of the GPU is not reached. Motivated by the architecture changes made by the GA3C algorithm, which gave A3C better GPU acceleration, together with the high learning efficiency of the UNREAL algorithm, this paper extends GA3C with the auxiliary tasks from UNREAL to create a Deep Reinforcement Learning algorithm, GUNREAL, with higher learning efficiency and also benefiting from GPU acceleration. We show that our GUNREAL system reaches higher scores on several games in the same amount of time than GA3C.