Label-Based Alignment Multi-Source Domain Adaptation for Cross-Subject EEG Fatigue Mental State Evaluation.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Accurate detection of driving fatigue is vital for reducing road accidents.
• Electroencephalogram (EEG) is an effective tool for assessing mental fatigue.
• Cross-subject EEG fatigue evaluation is challenging due to non-linearity and individual differences.

Purpose of the Study:

• To propose a novel method for cross-subject EEG fatigue mental state evaluation.
• To address the challenges posed by individual differences in EEG data.
• To improve the accuracy and generalizability of EEG-based fatigue detection.

Main Methods:

• Developed Label-based Alignment Multi-Source Domain Adaptation (LA-MSDA).
• LA-MSDA aligns label-based feature distributions to mitigate individual differences.
• Incorporated global optimization to refine classifier decision boundaries and enhance generalization.

Main Results:

• LA-MSDA demonstrated remarkable performance in cross-subject EEG fatigue evaluation.
• The method effectively eliminates the negative impact of significant individual differences.
• Achieved improved generalization ability for EEG fatigue detection.

Conclusions:

• LA-MSDA offers a promising solution for cross-subject EEG fatigue evaluation.
• The proposed method has wide application prospects in brain-computer interaction (BCI).
• Potential applications include online driver fatigue monitoring and advanced safety systems.