An Effective Fuzzy Support Vector Machine based Seizure Detection Model using EEG Signals

Abstract

Recently, the identification of the recorded epileptic seizure activities in Electroencephalography (EEG) signals becomes essential to classify seizures. Since manual detection of seizure is a tedious and difficult task for neurologists, an automated epilepsy detection model is demanded. Conventional EEG recognition techniques are solely based on artificial experience and weak generalization capability. For resolving these issues, this paper presents an automated seizure detection model using the Fuzzy Support Vector Machine (FSVM) model. HSVM is developed for multi-label classification where a region with related membership function is available for every class label and a data point is categorized to a multi-label class whose membership function is the higher. The performance of the FSVM model has been validated against the EEG signal dataset and the results are examined under several measures. The obtained results depicted that the FSVM model has achieved latter results with the maximum precision of 79.45%, recall of 79.63%, F-score of 78.50%, and accuracy of 79.65%.