Frequency-Domain Hybrid Model for EEG-Based Emotion Recognition

Emotion recognition based on Electroencephalogram (EEG) signals plays a vital role in affective computing and human–computer interaction (HCI). However, noise, artifacts, and signal distortions present challenges that limit classification accuracy and robustness. To address these issues, we propose ECA-ResDNN, a novel hybrid model designed to leverage the frequency, spatial, and temporal characteristics of EEG signals for improved emotion recognition. Unlike conventional models, ECA-ResDNN integrates an Efficient Channel Attention (ECA) mechanism within a residual neural network to enhance feature selection in the frequency domain while preserving essential spatial information. A Deep Neural Network further extracts temporal dependencies, improving classification precision. Additionally, a hybrid loss function that combines cross-entropy loss and fuzzy set loss enhances the model’s robustness to noise and uncertainty. Experimental results demonstrate that ECA-ResDNN significantly outperforms existing approaches in both accuracy and robustness, underscoring its potential for applications in affective computing, mental health monitoring, and intelligent human–computer interaction.