An Efficient Deep Learning based Models for Epileptic Seizure Detection using EEG data

Abstract

Machine learning algorithms, excluding deep learning algorithms, have been proposed to address the seizure prediction problem. Since EEG signals vary across patients due to differences in seizure type and location, most seizure prediction methods are specific to each patient. These algorithms employ various techniques for extracting, selecting, and classifying EEG features. However, a significant drawback of these methods is their reliance on manually extracted features, making it difficult to determine the most informative features that accurately represent each class. In a more recent trend, seizure prediction algorithms based on deep learning are employed, which integrate feature extraction and classification stages into a single automated framework. The objective of this paper is to develop deep learning-based algorithms for automatic feature learning, capable of being applied to all patients with minimal feature engineering and preprocessing requirements.