Complete Multi-Domain Decoupled Fusion Model for EEG-Based Person Identification
Zhixun Wang, Jiayu Lu, Tianyang Liu, Ziteng Zhu, Xiaofeng Liu, Bin Wang
AI summary
Problem
Existing EEG-based person identification methods suffer from coupled cross-domain features, insufficient multi-domain fusion, and fixed-scale extractors that miss individual-specific multi-scale information.
Approach
The model decouples temporal, spatial, and spectral features using independent attention mechanisms, fuses them comprehensively across all domains, and employs an input-dependent adaptive multi-scale CNN to dynamically adjust feature extraction scales per individual.
Key results
- Independent temporal-spatial-spectral attention eliminates cross-domain parameter coupling
- Full-domain fusion mechanism integrates single, dual, and triple domain features
- Input-dependent adaptive multi-scale CNN dynamically adjusts kernel contributions
- Outperforms all state-of-the-art methods across four EEG datasets
Why it matters
Advances reliable EEG-based biometric authentication for secure identity verification by capturing highly individual-specific neural patterns.
Abstract
Electroencephalogram (EEG) signals have unique individual characteristics and have broad application prospects in identity authentication. At present, person identification (PI) based on EEG using the temporal-spatial-spectral feature ex- traction framework has achieved remarkable success. However, the existing methods suffer from coupled cross-domain feature parameters and insufficient feature fusion during feature ex- traction, which limits the recognition ability. Moreover, fixed- scale feature extractors can hardly exploit the subject-specific multi-scale information. To address these challenges, we propose CMDFM: a complete multi-domain decoupled fusion model for EEG-based PI. Firstly, we design an independent temporal- spatial-spectral attention mechanism to eliminate cross-domain parameter coupling. Secondly, a full-domain fusion mechanism is designed to comprehensively integrate the features of the tem- poral domain, spatial domain and spectral domain. Finally, an adaptive multi-scale CNN is designed to adjust the contribution of the multi-scale convolution kernel, thereby making full use of individual-specific multi-scale information. We use four datasets to verify our method. The experimental results show that our method is superior to all the state-of-the-art methods. The code of CMDFM is at https://github.com/2538441690/CMDFM.