GCMCG: A Clustering-Aware Graph Attention and Expert Fusion Network for Multi-Paradigm, Multi-task, and Cross-Subject EEG Decoding

2025-11-29

AI Summary

Addressing challenges in EEG decoding for motor execution (ME) and motor imagery (MI)—including complex spatiotemporal dynamics, low signal-to-noise ratio, and poor cross-subject/cross-paradigm generalizability—this paper proposes a unified decoding framework integrating graph attention mechanisms with a mixture of experts. Methodologically, it introduces a pretrainable graph tokenization module to explicitly model electrode topological relationships and incorporates unsupervised spectral clustering to partition functionally coherent brain regions, enhancing interpretability. The framework integrates ICA-wavelet denoising, graph attention networks (GAT), and complementary CNN-GRU expert subnetworks, fused via a learnable gating mechanism and regularized with L1 penalty; training employs focal loss and a three-stage progressive sampling strategy. Evaluated on three public datasets, the method achieves accuracies of 86.60%, 98.57%, and 99.61%, respectively, demonstrating substantial improvements in cross-subject generalization and practical BCI performance.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Humans and AI: Brain-Sensing and Analysis

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Brain-Computer Interfaces (BCIs) based on Motor Execution (ME) and Motor Imagery (MI) electroencephalogram (EEG) signals offer a direct pathway for human-machine interaction. However, developing robust decoding models remains challenging due to the complex spatio-temporal dynamics of EEG, its low signal-to-noise ratio, and the limited generalizability of many existing approaches across subjects and paradigms. To address these issues, this paper proposes Graph-guided Clustering Mixture-of-Experts CNN-GRU (GCMCG), a novel unified framework for MI-ME EEG decoding. Our approach integrates a robust preprocessing stage using Independent Component Analysis and Wavelet Transform (ICA-WT) for effective denoising. We further introduce a pre-trainable graph tokenization module that dynamically models electrode relationships via a Graph Attention Network (GAT), followed by unsupervised spectral clustering to decompose signals into interpretable functional brain regions. Each region is processed by a dedicated CNN-GRU expert network, and a gated fusion mechanism with L1 regularization adaptively combines these local features with a global expert. This Mixture-of-Experts (MoE) design enables deep spatio-temporal fusion and enhances representational capacity. A three-stage training strategy incorporating focal loss and progressive sampling is employed to improve cross-subject generalization and handle class imbalance. Evaluated on three public datasets of varying complexity (EEGmmidb-BCI2000, BCI-IV 2a, and M3CV), GCMCG achieves overall accuracies of 86.60%, 98.57%, and 99.61%, respectively, which demonstrates its superior effectiveness and strong generalization capability for practical BCI applications.

Problem

• Develops a unified EEG decoding framework for motor execution and imagery
• Addresses EEG's low signal-to-noise ratio and cross-subject generalization issues
• Enhances spatio-temporal feature fusion via graph attention and clustering

Innovation

• Graph Attention Network models electrode relationships dynamically
• Unsupervised spectral clustering decomposes signals into functional brain regions
• Mixture-of-Experts CNN-GRU with gated fusion combines local and global features