Convolution spatial-temporal attention network for EEG emotion recognition

Area of Science:

• Neuroscience
• Artificial Intelligence
• Biomedical Engineering

Background:

• Emotion recognition from electroencephalogram (EEG) signals is a growing field due to its non-invasive nature and high temporal resolution.
• Traditional methods often rely on manual feature engineering, which can be time-consuming and may not capture the full complexity of EEG data.

Purpose of the Study:

• To introduce a novel, data-driven deep learning method for emotion recognition using EEG signals.
• To bypass manual feature engineering by transforming EEG signals into 3D spatio-temporal representations.

Main Methods:

• EEG signals were preprocessed and transformed from 2D time sequences into 3D spatio-temporal representations, emphasizing topological relationships between channels.
• A deep learning model combining convolutional neural networks (CNNs) and attention mechanisms was employed for automatic feature extraction and learning inter-channel dependencies.

Main Results:

• The proposed method achieved high accuracy in recognizing emotional states: 98.62% for arousal and 98.47% for valence.
• These results significantly surpass previous state-of-the-art performances (95.76% for arousal, 95.15% for valence).

Conclusions:

• The developed approach demonstrates the effectiveness of leveraging 3D spatio-temporal representations and deep learning for robust emotion recognition from EEG.
• The findings open new avenues for research in advanced emotion recognition technologies.