Identification of Epileptic EEG Signals Through TSK Transfer Learning Fuzzy System

Area of Science:

• Biomedical Engineering
• Machine Learning
• Signal Processing

Background:

• Epilepsy diagnosis relies heavily on electroencephalogram (EEG) signal analysis.
• Current intelligent recognition methods for EEG signals face limitations such as strict data distribution requirements, neglect of intra-class information, and lack of interpretability.
• Addressing these challenges is crucial for advancing automated epilepsy detection.

Purpose of the Study:

• To propose a novel, interpretable model for identifying epilepsy EEG signals.
• To overcome the limitations of existing methods, particularly in scenarios with data drift.
• To enhance the effectiveness of transfer learning in EEG signal recognition.

Main Methods:

• Development of a Takagi-Sugeno-Kang (TSK) transfer learning fuzzy system (TSK-TL).
• Integration of transfer learning principles with an interpretable TSK fuzzy system.
• Relaxation of data distribution requirements by effectively utilizing source and target domain information.

Main Results:

• The proposed TSK-TL model demonstrates interpretability.
• The model effectively identifies epilepsy EEG signals even under data drift conditions.
• Experimental results show superior performance of TSK-TL compared to existing algorithms in epilepsy EEG recognition.

Conclusions:

• The TSK-TL model offers a promising solution for interpretable and robust epilepsy EEG signal identification.
• This approach enhances the applicability of intelligent recognition technologies in real-world clinical settings with varying data distributions.
• The findings suggest a significant advancement in the field of automated EEG analysis for epilepsy detection.