An Interpretable Deep Learning Classifier for Epileptic Seizure Prediction Using EEG Data

Deep learning has served pattern classification in many applications, with a performance which often well exceeds that of other machine paradigms. Yet, general, deep used computational architectures built, albeit partially, by ad hoc means, and its decisions are not necessarily interpretable terms knowledge relevant to the application it serves. This is referred as black box problem, certain such epileptic seizure prediction, can be serious impediment. The purpose this study investigate an classifier for EEG-driven prediction. neural network because layers visualized interpreted result novel architecture where learned weights follow from signal processing computations frequency sub-band spatial filters. Consequently, extracted features no longer abstract they correspond commonly decoding EEG data. In addition, uses layer-wise relevance propagation reveal pertinent further explain leading decisions. prediction experiments using CHB-MIT data set, method produced results improved on state-of-the art, first layer filters corresponding clinically bands, input channels brain location originates contributing most significantly predictions.