Enhancing Emotion-Brain Representations With Orthogonal Fuzzy Power-Coherence Alignment.

Area of Science:

• Affective computing
• Human-computer interaction
• Neuroscience

Background:

• Electroencephalogram (EEG)-based emotion recognition faces challenges due to emotional fuzziness and complex EEG signals.
• Existing methods struggle to capture the nuanced nature of emotions and the intricate patterns within EEG data.

Purpose of the Study:

• To propose a novel framework, Orthogonal Fuzzy Power-Coherence Alignment (OFPCA), for enhanced emotional brain representation learning.
• To improve the accuracy and robustness of EEG-based emotion recognition.

Main Methods:

• OFPCA utilizes a Takagi-Sugeno-Kang (TSK) fuzzy system to model the fuzzy nature of emotions.
• It employs contrastive learning to align local-energy (power spectral density) and cross-regional interaction (coherence) views of EEG signals.
• Orthogonal regularization loss is incorporated for robust representation learning.

Main Results:

• OFPCA achieved high accuracy in cross-subject settings (41.14% on CRED, 83.24% on SEED) and intra-subject settings (63.24% on CRED, 85.37% on SEED).
• Performance surpassed existing methods on benchmark datasets (SEED and CRED).

Conclusions:

• Leveraging the inherent fuzziness of emotions enhances emotion recognition capabilities.
• OFPCA presents a novel fuzzy learning-based framework for effective multi-view EEG learning in emotion recognition.