Combining Event-Related Energy with Competitive Clustering Nets to Classify EEG Signals

Brain-computer interfaces provide a communication channel between human neurons and machines using electroencephalography (EEG) signals. In this paper, an analysis system is presented for single-trial electroencephalogram (EEG) classification. Combined with event-related synchronization (ERS) and event-related desynchronization (ERD), the Competitive Hopfield Clustering Network (CHCN) is applied to classify event-related Energy, clustering the features of left and right hand movement in EEG signals. The ERS and ERD were analyzed, and energy values of the EEG produced during the event period were calculated and used as event features. Experimental results show the CHCN achieves promising results in clustering the features of left and right hand movement in EEG signals.