EEG-FRM: a neural network based familiar and unfamiliar face EEG recognition method

Area of Science:

• Neuroscience
• Computer Science
• Biometrics

Background:

• Familiar face recognition is crucial in fields like medicine, law enforcement, and deception detection.
• Existing methods for face recognition often rely on visual data, limiting their application in certain scenarios.

Purpose of the Study:

• To develop and evaluate a novel neural network-based method for cross-subject familiar and unfamiliar face recognition using electroencephalography (EEG) data.
• To investigate the neural correlates, specifically P300 responses, associated with familiar face recognition.

Main Methods:

• A Complex Trial Protocol was used to collect EEG data from 147 subjects during a familiar and unfamiliar face recognition experiment.
• A novel EEG-based Face Recognition Model (EEG-FRM) was proposed, integrating a multi-scale convolutional classification network with a maximum probability mechanism.
• The model incorporates an attention module and supervised contrastive learning to enhance classification performance.

Main Results:

• Familiar face stimuli elicited significant P300 event-related potentials, primarily in the parietal lobe.
• The proposed EEG-FRM achieved a balanced accuracy of 85.64%, a true positive rate of 73.23%, and a false positive rate of 1.96%.
• The model demonstrated superior performance compared to other existing methods.

Conclusions:

• The developed EEG-FRM is effective for cross-subject familiar/unfamiliar face recognition.
• EEG-based face recognition, particularly leveraging P300 responses, offers a viable alternative to traditional visual methods.
• The findings highlight the potential of the proposed model in diverse applications requiring reliable individual identification.