introducing a convolutional neural network and visualization of its filters for classification of eeg signal for ssvep task

Purpose: brain-computer interface (bci) systems are able to understand and execute commands through processing brain signals. it has numerous applications in the field of biomedical engineering such as rehabilitation, biometric and entertainment. a bci system consists of four major parts: signal acquisition, signal pre-processing, feature extraction and classification. steady state visually evoked potentials (ssvep) is one of the most common paradigms in bci systems, which is generally a response to visual stimuli with the frequency between 5 to 60 hz.materials and methods: in this study, we suggest a convolutional neural network (cnn) based model for the classification of eeg signal during ssvep task. for the evaluation, the model was tested with different channels and electrodes.results: results show that channels number 138 and 139 have the great potential to appropriately classify eeg signal.conclusion: using the suggested model and the mentioned channels, the accuracy of 73.74% could be achieved in this study.