Data Mining and Machine Learning Applications

15Classification of EEG Signals for Detection of Epileptic Seizure Using Restricted Boltzmann Machine Classifier

Sudesh Kumar, Rekh Ram Janghel* and Satya Prakash Sahu

National Institute of Technology, Raipur, India

Abstract

Epilepsy is a disease that is an electrophysiological disorder related to the brain and is characterized by various types of recurrent seizures. Electroencephalogram (EEG) is a test that is developed by various neurologists to capture the electrical signals that occur in the brain and is widely used for the Analysis and detection of epileptic seizures. As we know that it is tough to identify the various types of electrical activities by visual inspection; thus, it opens up the vast research in the field of biomedical engineering to develop a system and various algorithms for the identification of these activities and changes in the human brain. Therefore an automated seizure detection system is needed for the classification of epileptic seizures. We handled the EEG dataset of CHB-MIT (scalp EEG) to discover if our model could outflank the best in class proposed models. We have proposed a methodology based on the Restricted Boltzmann Machine (RBM) neural network model, which is used to perform classification over the EEG signals among binary classes, namely a healthy (non-seizure) and non-healthy (seizure) classes. The analysis is performed on an open accessible CHB-MIT data set. The model performance is assessed based on various performance metrices like ...