Information transfer BMI by extracting common features among many users and situations

2017 Fiscal Year Final Research Report

• Project/Area Number: 15H02759
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Single-year Grants
• Section: 一般
• Research Field: Soft computing
• Research Institution: Advanced Telecommunications Research Institute International
• Principal Investigator: KAWANABE Motoaki 株式会社国際電気通信基礎技術研究所, 脳情報通信総合研究所, 研究室長 (30272389)
• Co-Investigator(Kenkyū-buntansha): 兼村 厚範 国立研究開発法人産業技術総合研究所, 情報・人間工学領域, 研究員 (50580297) ; 平山 淳一郎 株式会社国際電気通信基礎技術研究所, 脳情報通信総合研究所, 研究員 (80512269)
• Co-Investigator(Renkei-kenkyūsha): HIRAYAMA Jun-ichiro 株式会社国際電気通信基礎技術研究所, 脳情報通信総合研究所, 研究員 (80512269)
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: 確率的情報処理 / 機械学習 / ブレインマシンインタフェース

Outline of Final Research Achievements

The brain activity patterns during same cognitive tasks differ between users, and vary depending on situations even within each single user. We developed analysis methods for understanding such heterogeneity of brain measurement data and applied them to real fMRI and EEG data. In order to find common features among users and situations, we also evaluated the noise removal methods that are useful for alleviating fluctuations and constructed a procedure for robust feature extraction. Furthermore, we proposed frameworks of transfer learning based on dictionary learning and multivariate autoregression models towards brain machine interface (BMI) with lower users' burden. We tested their usefulness by using 30 days myoelectric data.

Free Research Field

医用工学・機械学習