Multi-class motor imagery classification by singular value decomposition and deep boltzmann machine

Abstract

Motor function rehabilitation is very urgent for patients. Motor imagery is an efficient way for rehabilitation. To achieve the supervision of multiple rehabilitation targets simultaneously, the promotion of multi-class motor imagery classification accuracy is critical. In this paper, a multi-class classification method is proposed by utilizing singular value decomposition and deep boltzmann machine. Singular value decomposition is applied to suppress the artifacts and acquire the channel-individual characteristics. The deep boltzmann machine is employed to extract and model the characteristics and achieve the motor imagery classification. Results demonstrate that the proposed method has achieved a 14.2% higher classification accuracy than the common spatial pattern on average. This results are further validated by the statistical methods, which present a significant difference (p < 0.05). The proposed method is favorable for promoting the multi-class motor imagery classification efficiency.