Presentation | 2017-09-15 Fast and General-Purpose Bayesian Optimization using Tree-Based Model with Gaussian Process Hiroo Iwanaga, Yukio Ohsawa, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Bayesian optimization is an effective method for black-box optimization problems such as hyperparameter tuning of machine learning algorithms that require efficient search of various parameters. Gaussian process (GP) is widely used to estimate objective functions in the sequence of bayesian optimization, but to use GP in practical situations, there are some difficult points as follows; computational cost per training sample is cubic and GP cannot treat non-continuous parameters naturally. To apply bayesian optimization for practical tasks that contains discrete and conditional parameters, bayesian optimization method based on random forest has been proposed, but it causes poor estimation of variance in unsearched areas. In this work, we propose a bayesian optimization method using tree-based model (decision tree or random forest) combined with Gaussian process. We show stable performance of our method for optimization tasks that contain various type of parameters, while cutting computational cost down. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Bayesian Optimization / Random Forest / Decision Tree / Gaussian Process |
Paper # | PRMU2017-48,IBISML2017-20 |
Date of Issue | 2017-09-08 (PRMU, IBISML) |
Conference Information | |
Committee | PRMU / IBISML / IPSJ-CVIM |
---|---|
Conference Date | 2017/9/15(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Shinichi Sato(NII) / Kenji Fukumizu(ISM) |
Vice Chair | Hironobu Fujiyoshi(Chubu Univ.) / Yoshihisa Ijiri(Omron) / Masashi Sugiyama(Univ. of Tokyo) |
Secretary | Hironobu Fujiyoshi(AIST) / Yoshihisa Ijiri(NAIST) / Masashi Sugiyama(Kyoto Univ.) / (Univ. of Tokyo) |
Assistant | Masato Ishii(NEC) / Yusuke Sugano(Osaka Univ.) / Ichiro Takeuchi(Nagoya Inst. of Tech.) / Toshihiro Kamishima(AIST) |
Paper Information | |
Registration To | Technical Committee on Pattern Recognition and Media Understanding / Technical Committee on Infomation-Based Induction Sciences and Machine Learning / Special Interest Group on Computer Vision and Image Media |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Fast and General-Purpose Bayesian Optimization using Tree-Based Model with Gaussian Process |
Sub Title (in English) | |
Keyword(1) | Bayesian Optimization |
Keyword(2) | Random Forest |
Keyword(3) | Decision Tree |
Keyword(4) | Gaussian Process |
1st Author's Name | Hiroo Iwanaga |
1st Author's Affiliation | The University of Tokyo/NTT DATA Mathematical Systems Inc.(Univ. of Tokyo/NTT DATA MSI) |
2nd Author's Name | Yukio Ohsawa |
2nd Author's Affiliation | The University of Tokyo(Univ. of Tokyo) |
Date | 2017-09-15 |
Paper # | PRMU2017-48,IBISML2017-20 |
Volume (vol) | vol.117 |
Number (no) | PRMU-210,IBISML-211 |
Page | pp.pp.67-74(PRMU), pp.67-74(IBISML), |
#Pages | 8 |
Date of Issue | 2017-09-08 (PRMU, IBISML) |