Epileptic Seizure Detection using Two-Layer Feature Extraction and Hyper-Parameter Optimization

Epileptic seizures happen owing to anarchy in intellect functionality that can influence patient's physical condition. Finding of epileptic seizures inception is fairly valuable for medication and emergency alerts. Machine learning techniques and computational methods play a key part in detecting epileptic seizures from Electroencephalograms (EEG) signals. The main objective of this work is to provide an ANN framework with optimized performance related to seizure detection. Here, a machine learning framework is employed for seizure detection where the two-layer feature extraction with ANN classifiers are used to categorize seizure and non-seizure data. To get better performance, the best parameters related to ANN with the dataset are identified through bayes-optimization method. This model affords a trustworthy feature extraction and optimization for training a detection model. The proposed model is evaluated using the popular public dataset CHB-MIT.