A LightGBM-Based EEG Analysis Method for Driver Mental States Classification.

Area of Science:

• Neuroscience
• Computer Science
• Transportation Safety

Background:

• Driver fatigue is a significant cause of road traffic accidents.
• Electroencephalography (EEG) is increasingly used to monitor physiological and brain activity for fatigue detection.
• Efficient and timely detection of driver mental states using EEG remains a challenge.

Purpose of the Study:

• To develop a novel, light-weighted classifier for EEG-based mental state identification in drivers.
• To evaluate the performance of the proposed classifier against traditional machine learning models.
• To assess the transfer learning capabilities of the new model for driver mental state classification.

Main Methods:

• Utilized Common Spatial Pattern (CSP) for feature extraction from EEG data.
• Developed a light-weighted classifier named LightFD, based on the gradient boosting framework.
• Compared LightFD with Support Vector Machine (SVM), Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU), and Large Margin Nearest Neighbor (LMNN).

Main Results:

• The LightFD classifier demonstrated superior classification performance and decision efficiency compared to traditional models.
• LightFD exhibited enhanced transfer learning performance in classifying driver mental states from EEG.
• The proposed model achieved comparable results to established classifiers while being more efficient.

Conclusions:

• The developed LightFD classifier offers improved real-time EEG mental state prediction capabilities.
• LightFD shows significant potential for practical applications in brain-computer interaction (BCI) for driver monitoring.
• This approach contributes to enhancing road safety by providing a more effective tool for fatigue detection.