Related Experiment Videos
CRGFFNet: an adaptive inter-channel relation guided feature fusion network for enhanced EEG personality recognition
Bin Yan1, Tong Zhang1,2, Zhengxiu Li1,2
1School of Information Engineering, Nanchang Hangkong University, Nanchang 330000, People's Republic of China.
Biomedical Physics & Engineering Express
|July 15, 2026
Summary
This study introduces a novel network for electroencephalography (EEG)-based personality recognition, improving accuracy by effectively fusing multi-dimensional features and preserving spatial topology for better clinical psychology applications.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Psychology
Background:
- Personality recognition using electroencephalography (EEG) shows promise but is limited by poor spatial topology exploitation and ineffective multi-dimensional feature fusion.
- Existing methods struggle to capture the complex relationships within EEG signals for accurate personality assessment.
Purpose of the Study:
- To propose an adaptive Inter-Channel Relation Guided Feature Fusion Network (CRGFFNet) to enhance EEG-based personality recognition.
- To address limitations in spatial topology and feature fusion for more accurate personality trait identification.
Main Methods:
- Developed the Spatial Proximal-Channel-Reordering (SPCR) algorithm using reinforcement learning to reorder EEG channels and reduce noise.
- Introduced the Spatio-Temporal-Frequency Hybrid Convolutional Feature Augmentation (STFA) module for extracting discriminative features across multiple signal domains.
- Utilized a Transformer encoder to model global contextual relationships from STFA module outputs.
Main Results:
- The CRGFFNet achieved state-of-the-art performance with an average accuracy of 89.94% across Big Five personality dimensions.
- The proposed model significantly outperformed existing methods in personality recognition tasks.
- The SPCR algorithm effectively preserved spatial structure and attenuated noise, while STFA enhanced feature extraction.
Conclusions:
- The CRGFFNet effectively overcomes limitations in EEG-based personality recognition by integrating spatial topology and multi-dimensional feature fusion.
- The proposed adaptive network architecture demonstrates superior performance, paving the way for advanced applications in clinical psychology and human-computer interaction.
