Lightweight Transformer for EEG Classification via Balanced Signed Graph Algorithm Unrolling

2025-10-03

AI Summary

This work addresses EEG signal classification for epilepsy patients versus healthy controls. We propose a lightweight, interpretable graph Transformer model. Methodologically, we (1) construct a balanced signed graph and introduce a similarity transformation to map negative-weight graph signals onto positive-weight graphs, unifying spectral-domain modeling and low-pass filtering; (2) implement ideal low-pass filtering on the transformed graph via Lanczos approximation and jointly learn the optimal cutoff frequency within an algorithm unrolling framework; and (3) integrate a dual-balanced graph denoising module into the Transformer architecture. The resulting model reduces parameter count by over 60% while matching the classification accuracy of state-of-the-art deep learning methods. It achieves a favorable trade-off among computational efficiency, model interpretability, and predictive performance.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Samples of brain signals collected by EEG sensors have inherent anti-correlations that are well modeled by negative edges in a finite graph. To differentiate epilepsy patients from healthy subjects using collected EEG signals, we build lightweight and interpretable transformer-like neural nets by unrolling a spectral denoising algorithm for signals on a balanced signed graph -- graph with no cycles of odd number of negative edges. A balanced signed graph has well-defined frequencies that map to a corresponding positive graph via similarity transform of the graph Laplacian matrices. We implement an ideal low-pass filter efficiently on the mapped positive graph via Lanczos approximation, where the optimal cutoff frequency is learned from data. Given that two balanced signed graph denoisers learn posterior probabilities of two different signal classes during training, we evaluate their reconstruction errors for binary classification of EEG signals. Experiments show that our method achieves classification performance comparable to representative deep learning schemes, while employing dramatically fewer parameters.

Problem

• Classifying epilepsy using EEG signals via lightweight transformers
• Modeling EEG anti-correlations with balanced signed graphs
• Achieving high accuracy with fewer parameters than deep learning

Innovation

• Unrolled spectral denoising algorithm for balanced signed graphs
• Implemented ideal low-pass filter via Lanczos approximation
• Built lightweight transformer-like neural nets for EEG classification