An Extreme Learning Machine Approach for Detection of Epilectic Seizure

Özet

Irregularities in the electrical activities of the brain can cause epileptic attacks. In the diagnosis and classification of epileptic attacks and epilepsy, data showing the electrical activity of the brain obtained by EEG has an important place. This study presents an EEG classification approach based on the extreme learning machine (ELM). The ELM algorithm is used for the characteristics of single hidden layer feedforward neural network (SLFN). EEG recordings of healthy individuals and individuals who had epileptic attacks were classified by using ELM. As a result of the study, the classification success of the method we applied was found to be 99.67%. It is predicted that the obtained system may be useful in evaluating the pre-diagnosis for physicians.