Estimating the Efficiency of Machine Learning Algorithms in Predicting Seizure With Convolutional Neural Network Architecture

Abstract

The reason trends in prevalent detection of EEG seizure help in analyzing the various features of EEG signals to customize and to remove visual inspection in reading the EEG signals. Epilepsy is a disorder and is identified by baseless seizures that have been associated with unexpected improper neural discharges which result in various health issues and also result in death. One of the most common methods in detecting contraction seizures is an electroencephalogram. By using machine learning methods, it is easy to extract the features of EEG signals that help in detecting seizures. Convolutional neural network (CNN) includes both inputs as well as output layers that help in training the data acquired since it helps in analyzing the large set of high dimensional data. The performance analysis is done under multiple classifiers such as random forest, gradient boosting, and decision tree, which are used in feature extraction. Among them, random forest proves to be the best classifier in achieving a high degree of accuracy.