Multi-class SVM for EEG signal classification using wavelet based approximate entropy

Abstract

In this paper, we have proposed a novel wavelet based approximate entropy for feature extraction and a novel Multi-Class Support Vector Machine (MSVM) for the multi-class electroencephalogram (EEG) signals classification with the emphasis on epileptic seizure detection. The aim was to determine an effective classifier and features for this problem. Wavelets have played an important role in biomedical signal processing for its ability to capture localized spatial-frequency information of EEG signals. The MSVM works well for high dimensional, multi-class data streams. Decision making was performed in two stages: feature extraction by computing the wavelet based approximate entropy and classification using the classifiers trained on the extracted features. We have compared the MSVM with Probabilistic Neural Network (PNN) by evaluating with the benchmark EEG dataset. Our experimental results show that the MSVM with wavelet based approximate entropy features gives high classification accuracies than the existing classifier. © Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2012.