Presentation 2022-01-18
Bayesian Optimization for Simultaneous Optimization of Multiple Tasks with Max-value Entropy Search
Rintaro Yamada, Shion Takeno, Masayuki Karasuyama,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Bayesian optimization (BO) has been widely studied as an effective approach to black-box optimizations. On the other hand, in some practical scenarios, multiple related optimization problems (tasks) need to be solved simultaneously instead of only solving single optimization problem. In this case, the optimization procedure can be accelerated by incorporating correlations among the optimization problems. We propose an information-theoretic acquisition function of BO for the multi-task setting, in which the information gains across different tasks are evaluated. In particular, our method is based on a BO methodcalled max-value entropy search (MES), by which we show that a natural extension for the multi-task setting can be derived. We also demonstrate effectiveness of the proposed method through empirical evaluations.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Bayesian optimization / multi-task / mutual information
Paper # IBISML2021-28
Date of Issue 2022-01-10 (IBISML)

Conference Information
Committee IBISML
Conference Date 2022/1/17(2days)
Place (in Japanese) (See Japanese page)
Place (in English) Online
Topics (in Japanese) (See Japanese page)
Topics (in English) Machine Learning, etc.
Chair Ichiro Takeuchi(Nagoya Inst. of Tech.)
Vice Chair Masashi Sugiyama(Univ. of Tokyo)
Secretary Masashi Sugiyama(Univ. of Tokyo)
Assistant Tomoharu Iwata(NTT) / Atsuyoshi Nakamura(Hokkaido Univ.)

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) Bayesian Optimization for Simultaneous Optimization of Multiple Tasks with Max-value Entropy Search
Sub Title (in English)
Keyword(1) Bayesian optimization
Keyword(2) multi-task
Keyword(3) mutual information
1st Author's Name Rintaro Yamada
1st Author's Affiliation Nagoya Institute of Technology(NIT)
2nd Author's Name Shion Takeno
2nd Author's Affiliation Nagoya Institute of Technology(NIT)
3rd Author's Name Masayuki Karasuyama
3rd Author's Affiliation Nagoya Institute of Technology(NIT)
Date 2022-01-18
Paper # IBISML2021-28
Volume (vol) vol.121
Number (no) IBISML-321
Page pp.pp.75-80(IBISML),
#Pages 6
Date of Issue 2022-01-10 (IBISML)