EEG-Based Emotion Monitoring and Regulation System by Learning the Discriminative Brain Network Manifold.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Signal Processing

Background:

• Emotion recognition using electroencephalogram (EEG) is crucial for human-like artificial intelligence.
• EEG signals are susceptible to noise and individual variability, complicating emotion-dependent pattern extraction.
• Existing methods struggle with noise suppression and accurate emotion classification from EEG.

Purpose of the Study:

• To propose a novel L1-norm space defined discriminative brain network manifold learning model (L1-SGL) for robust EEG-based emotion recognition.
• To effectively separate EEG noise outliers and automatically correct pseudolabeled samples.
• To implement and validate an online system for real-time emotion monitoring and regulation.

Main Methods:

• Development of the L1-norm space defined discriminative brain network manifold learning model (L1-SGL).
• Utilizing L1-norm minimization for noise outlier separation and pseudolabel correction.
• Offline and online experimental validation with EEG data from human participants.

Main Results:

• The L1-SGL model demonstrated superior performance in EEG emotion recognition compared to existing methods.
• Offline experiments showed effective noise suppression and improved pattern extraction.
• Online emotion decoding achieved 86.30% accuracy, confirming real-time feasibility.
• Significant results in negative emotion regulation (p < 0.001) confirmed the model's effectiveness.

Conclusions:

• The L1-SGL model offers a promising solution for accurate and robust EEG-based emotion recognition.
• The model's time efficiency enables real-time applications like affective brain-computer interfaces (aBCIs).
• The study validates the feasibility of L1-SGL for intelligent clinical closed-loop treatments and interactive applications.