The Hybrid Classification Model Thanks to Artificial Neural Network and Artificial Immune Systems for Diagnosis of Epilepsy from Electroencephalography

Abstract

In this study, Artificial Neural Networks (ANN) and Artificial Immune (AI) techniques designed in the form of a hybrid structure are used for diagnosis of epilepsy patients via EEG signals. Attributes of EEG signals are needed to be determined by employing EEG signals which are recorded using EEG. In this process the raw digital signals data is received and is summarized in some respects. From this data, four characteristics are extracted for the classification process. 20% of available data is reserved for testing while 80% of available data is being reserved for training. These actions were repeated five times by performing cross-validation process. AIS is used for updating the weights during training ANN and a program is constituted for the classification of EEG signals. Education and recording processes were performed with different parameters by means of the constituted program. The obtained findings show that the proposed method was effective for achieving accurate results as much as possible with the use of ANN and AIS, together.