Presentation | 2020-06-29 A Study on Model-based Deep Reinforcement Learning Using Autonomous Search for Subgoal Motoki Maruyama, Satoshi Endo, Koji Yamada, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Model-based deep reinforcement learning (DRL) is more sample-efficient than model-free DRL. But it requires a deep generative model to learn an accurate dynamics. Therefore, it is difficult to deep lookahead depth due to a realistic cost. In this work, We propose to compare the subgoals with the shallow lookahead depth and give rewards according to the proximity by decomposing the tasks and setting subgoals. This method achieve a certain result in the maze. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Deep reinforcement learning / Subgoals / Model-based / DQN / GANs |
Paper # | NC2020-6,IBISML2020-6 |
Date of Issue | 2020-06-22 (NC, IBISML) |
Conference Information | |
Committee | NC / IBISML / IPSJ-BIO / IPSJ-MPS |
---|---|
Conference Date | 2020/6/29(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Online |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Kazuyuki Samejima(Tamagawa Univ) / Ichiro Takeuchi(Nagoya Inst. of Tech.) / Hiroyuki Kurata(Kyutech) / Masakazu Sekijima(Tokyo Tech) |
Vice Chair | Rieko Osu(Waseda Univ.) / Masashi Sugiyama(Univ. of Tokyo) / Koji Tsuda(Univ. of Tokyo) |
Secretary | Rieko Osu(NTT) / Masashi Sugiyama(ATR) / Koji Tsuda(AIST) / (NTT) / (Chuo Univ.) |
Assistant | Ken Takiyama(TUAT) / Nobuhiko Wagatsuma(Toho Univ.) / Atsuyoshi Nakamura(Hokkaido Univ.) / Shigeyuki Oba(Kyoto Univ.) |
Paper Information | |
Registration To | Technical Committee on Neurocomputing / Technical Committee on Infomation-Based Induction Sciences and Machine Learning / IPSJ Special Interest Group on Bioinformatics and Genomics / IPSJ Special Interest Group on Mathematical Modeling and Problem Solving |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Study on Model-based Deep Reinforcement Learning Using Autonomous Search for Subgoal |
Sub Title (in English) | |
Keyword(1) | Deep reinforcement learning |
Keyword(2) | Subgoals |
Keyword(3) | Model-based |
Keyword(4) | DQN |
Keyword(5) | GANs |
1st Author's Name | Motoki Maruyama |
1st Author's Affiliation | University of the Ryukyus(Univ. of the Ryukyus) |
2nd Author's Name | Satoshi Endo |
2nd Author's Affiliation | University of the Ryukyus(Univ. of the Ryukyus) |
3rd Author's Name | Koji Yamada |
3rd Author's Affiliation | University of the Ryukyus(Univ. of the Ryukyus) |
Date | 2020-06-29 |
Paper # | NC2020-6,IBISML2020-6 |
Volume (vol) | vol.120 |
Number (no) | NC-79,IBISML-80 |
Page | pp.pp.33-38(NC), pp.33-38(IBISML), |
#Pages | 6 |
Date of Issue | 2020-06-22 (NC, IBISML) |