Presentation 2022-12-22
Implementation and evaluation of NoisyNets to Reinforcement learning of Automated Designing ICT System
Tianchen Zhou, Yutaka Yakuwa, Natsuki Okamura, Takayuki Kuroda, Ikuko E. Yairi,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper introduces a reinforcement learning method for the ICT system design process. Since the state space of the designing process is complex and rewards obtained in the design process are significantly sparse. The proposed method to apply an additional noisy layer to the structure of the graph neural network for reinforcement learning in an automated design technology for ICT systems called Weaver. The parametric noise applied to the network are learned with gradient descent along with the remaining network weights. This helps to reduce harmful overestimation from the network and aids the efficiency of exploration for Weaver. The evaluation result showed that using the proposed algorithm for Weaver could shorten the time to learn the design of ICT systems.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) System Design / Design Automation / Machine Learning / Reinforcement Learning
Paper # IBISML2022-50
Date of Issue 2022-12-15 (IBISML)

Conference Information
Committee IBISML
Conference Date 2022/12/22(2days)
Place (in Japanese) (See Japanese page)
Place (in English) Kyoto University
Topics (in Japanese) (See Japanese page)
Topics (in English) Machine Learning, etc.
Chair Masashi Sugiyama(Univ. of Tokyo)
Vice Chair Toshihiro Kamishima(AIST) / Koji Tsuda(Univ. of Tokyo)
Secretary Toshihiro Kamishima(NTT) / Koji Tsuda(Hokkaido Univ.)
Assistant Yoshinobu Kawahara(Osaka Univ.) / Taiji Suzuki(Tokyo Inst. of Tech.)

Paper Information
Registration To Technical Committee on Infomation-Based Induction Sciences and Machine Learning
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Implementation and evaluation of NoisyNets to Reinforcement learning of Automated Designing ICT System
Sub Title (in English)
Keyword(1) System Design
Keyword(2) Design Automation
Keyword(3) Machine Learning
Keyword(4) Reinforcement Learning
1st Author's Name Tianchen Zhou
1st Author's Affiliation Sophia University(Sophia Univ.)
2nd Author's Name Yutaka Yakuwa
2nd Author's Affiliation NEC Corporation(NEC)
3rd Author's Name Natsuki Okamura
3rd Author's Affiliation Sophia University(Sophia Univ.)
4th Author's Name Takayuki Kuroda
4th Author's Affiliation NEC Corporation(NEC)
5th Author's Name Ikuko E. Yairi
5th Author's Affiliation Sophia University(Sophia Univ.)
Date 2022-12-22
Paper # IBISML2022-50
Volume (vol) vol.122
Number (no) IBISML-325
Page pp.pp.46-53(IBISML),
#Pages 8
Date of Issue 2022-12-15 (IBISML)