Enhancing cross-subject emotion recognition precision through unimodal EEG: a novel emotion preceptor model

Area of Science:

• Affective computing
• Neuroscience
• Computer Science
• Psychology

Background:

• Emotion recognition technology is advancing, with physiological signals like electroencephalogram (EEG) showing promise.
• Individual differences in EEG signals create noise, hindering accurate emotion recognition.
• Multimodal data collection for EEG poses practical challenges due to equipment and environmental constraints.

Purpose of the Study:

• To develop a cross-subject emotion recognition model using unimodal EEG signals.
• To overcome limitations of individual differences and multimodal data collection in EEG-based emotion recognition.
• To enhance the accuracy and practical applicability of affective computing.

Main Methods:

• Proposed the Emotion Preceptor, a model utilizing unimodal EEG signals for cross-subject emotion recognition.
• Introduced a Static Spatial Adapter to integrate spatial information and mitigate individual differences in EEG data.
• Employed a Temporal Causal Network to extract relevant temporal features for precise emotion recognition.

Main Results:

• The Emotion Preceptor demonstrated superior performance on the SEED and SEED-V datasets.
• Validated a novel data processing method combining Differential Entropy (DE) features in a temporal sequence.
• Analyzed model performance through biological interpretability and neuroscience research.

Conclusions:

• The Emotion Preceptor effectively achieves precise emotion recognition using unimodal EEG signals.
• The proposed methods reduce individual variability and improve the robustness of emotion recognition.
• This research advances EEG-based emotion recognition and affective computing applications.