A Brain Biometric-based Identification Approach Using Local Field Potentials.

Area of Science:

• Neuroscience
• Biometrics
• Machine Learning

Background:

• Traditional biometrics (face, iris, fingerprint) face privacy and security challenges.
• Brain signal identification offers unique advantages like confidentiality, anti-spoofing, continuity, and cancelability.
• Local field potentials (LFPs) are suitable for biometrics due to stability, signal quality, and resolution.

Purpose of the Study:

• To propose a novel, secure biometric system based on brain signals.
• To leverage deep neural networks for extracting unique identifiers from LFP signals.
• To develop a cancelable biometric that addresses privacy concerns.

Main Methods:

• Extraction of biometric features from local field potential (LFP) signals.
• Utilizing a deep neural network for biometric generation and identification.
• Designing the biometric to be task-related for cancelability.
• Conducting experiments on ten rats for performance evaluation.

Main Results:

• The proposed LFP-based biometric achieved a high identification accuracy of 94.47%.
• The system demonstrated stable performance over several days.
• The task-related generation ensures the biometric is cancelable.

Conclusions:

• Brain signal biometrics, specifically using LFPs and deep learning, present a secure and viable alternative to traditional methods.
• The developed cancelable biometric system effectively addresses privacy and anti-spoofing concerns.
• This approach shows significant promise for future biometric security applications.