Seizure Classification From EEG Signals Using Transfer Learning, Semi-Supervised Learning and TSK Fuzzy System.

Area of Science:

• * Computational neuroscience and biomedical engineering.
• * Development of advanced machine learning algorithms for clinical applications.

Background:

• * Accurate epileptic seizure recognition from electroencephalogram (EEG) signals is crucial for epilepsy diagnosis.
• * Current machine learning methods face challenges with data distribution discrepancies, insufficient training data, and lack of model interpretability.
• * Manual labeling of EEG data by clinicians is time-consuming and prone to inconsistency.

Purpose of the Study:

• * To improve the accuracy and interpretability of machine learning-based epileptic seizure classification.
• * To address limitations of existing methods, including data distribution differences and limited training data.
• * To develop a transparent and reliable system for automated seizure detection.

Main Methods:

• * Integration of transductive transfer learning to minimize data distribution discrepancies between training and testing sets.
• * Application of semi-supervised learning to leverage unlabeled testing data, augmenting limited training datasets.
• * Utilization of a TSK (Takagi-Sugeno-Kang) fuzzy system to enhance model interpretability.

Main Results:

• * The proposed integrated approach demonstrated superior performance compared to existing state-of-the-art seizure classification algorithms.
• * Transfer learning effectively reduced data distribution gaps, improving generalization.
• * Semi-supervised learning compensated for insufficient labeled training data, boosting classification accuracy.

Conclusions:

• * The combined use of transfer learning, semi-supervised learning, and TSK fuzzy systems offers a robust solution for epileptic seizure recognition.
• * The developed system achieves higher accuracy and provides greater interpretability than conventional methods.
• * This approach holds significant potential for enhancing clinical diagnosis and management of epilepsy.