P300 Detection Adaptive Channel Selection Method under the Multiple Kernel Learning

Abstract

This study presents an adaptive channel selection method for P300 detection. The proposed adaptive channel selection method selects the channel efficiently by introducing multiple kernel learning. The multiple kernel learning proposed in this paper selects the model, maps the EEG signals in different acquisition channels into different feature spaces through different kernel functions at first. Then it constructs multiple kernel functions by linear weighting and uses multiple kernel classifier training process to learn weight coefficients to adaptively select the optimal sampling set channel combination. The experiment demonstrates that the adaptive channel selection method under the multiple kernel learning is reasonable.