A comparison study on EEG signal classification using Component analysis (PCA, ICA) and Support Vector Machine (SVM)

Abstract

This studyaims to analyze and process Electroencephalogram (EEG)signals using an automated classification method with Support vector machine (SVM), to categorize patient’s seizure: epileptic or non-epileptic. We employed a framework of signal analysis techniques, and we started by applying discrete wavelet decomposition(DWT) on the original signal, followed by extracting a set of statistical features and building the feature matrix. Next, a feature reduction PCA and ICA were explored to represent the data in a new distinct space with reduced dimension. Finally, an SVM algorithm was trained and used upon a set of testing data to be classified: epileptic or not. The performance of classification process due to different methods is presented and compared to show the excellent classification process.