Extension of nonparametric Bayesian methods to semiparametric models and its applications

2015 Fiscal Year Final Research Report

• Project/Area Number: 25280083
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Partial Multi-year Fund
• Section: 一般
• Research Field: Intelligent informatics
• Research Institution: Nara Institute of Science and Technology
• Principal Investigator: Ikeda Kazushi 奈良先端科学技術大学院大学, 情報科学研究科, 教授 (10262552)
• Co-Investigator(Kenkyū-buntansha): SAKUMURA Yuichi 愛知県立大学, 情報科学部, 准教授 (50324968)
• Co-Investigator(Renkei-kenkyūsha): KUBO Takatomi 奈良先端科学技術大学院大学, 情報科学研究科, 特任准教授 (20631550) ; WATANABE Kazuho 豊橋技術科学大学, 工学研究科, 講師 (10506744)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: ノンパラメトリックベイズ法 / セミパラメトリックモデル / 運転行動モデリング / ヘビーテイル

Outline of Final Research Achievements

We tried to apply the semiparametric model to nonparametric Bayesian methods. However, we found that the estimation function method for the semiparametric model is not applicable to the wind data given by AMeDAS we treated in this study. Hence, we introduced CRPS or its extensions to the data, instead, and successfully predicted the wind speed for ten minutes to several hours horizons. In addition, we applied machine learning methods such as nonparametric Bayesian methods to driving data, motion capture data, web data, and mouse behavior data, and showed their effectiveness for modeling and prediction.

Free Research Field

機械学習