2013 Fiscal Year Annual Research Report
多重線型部分空間回帰法及び脳コンピュータインタフェースへの応用
| Project/Area Number |
24700154
|
|---|
| Research Institution | Institute of Physical and Chemical Research |
|---|
Principal Investigator |
ZHAO QIBIN 独立行政法人理化学研究所, 脳科学総合研究センター, 研究員 (30599618)
|
|---|
| Keywords | Machine learning / Tensor analysis / Kernel machines / Brain computer interface / Brain signal processing / Bayesian inference |
|---|
| Research Abstract |
(1) We have developed an online brain computer interface system for controlling robot arm and wheelchair by brain signals. Several paradigms have been implemented such as visual P300, auditory P300, hybrid P300 using visual and auditory stimuli.
(2) I have further improved the proposed tensor-based Gaussian processes methods by introducing the automatic learning of hyperparameters.
(3) We have successfully conducted EEG experiments with multi-subjects simultaneously; in addition, the movement trajectories of hands were also recorded. Two types of experiments including collaborative and competitions were performed.
(4) Aggregation of sparse linear discriminant analysis has been proposed and applied to EEG-based BCI system with significantly improved performance.
(5) I have applied newly proposed multilinear regression, and kernel-based tensor classifier to decoding of EEG/ECoG data, image and video analysis.
|
