A Semi-supervised Support Vector Machine Classification Method based on Parameter Optimization for a Motor Imagery based BCI System

Brain-computer interface (BCI) is a new channel for humans to communicate with the outside world. It can meet the needs of patients with severe neuromuscular disorders in their daily activities. How to construct BCI based on a small number of labeled samples is a key problem in the development of BCI. A semi-supervised classification model combining common space pattern for feature extraction and support vector machine (SVM) for classification can achieve better classification results under small sample conditions. On this basis, this paper proposes a new algorithm, which uses particle swarm optimization (PSO) to optimize the parameters of SVM, and constructs a semi-supervised classification model based on PSO and SVM to improve the classification effect of motor imagery under the condition of small samples. Experimental results show that the proposed method can achieve higher classification accuracies than the method without optimized parameters.