Abstract
Brain-computer interfaces have shown recent potential as a security tool in the field of biometric encryption. Most modern devices contain some form of biometric security such as facial recognition or fingerprint authorization however these methods have proven to be fallible. Brain data collected through electroencephalography (EEG) can be used in high-security applications to authenticate the correct user and inhibit fraudulent attempts at accessing sensitive data. The system designed in this research takes a different approach than other EEG encryption algorithms created in the past as the distributed nature of the system allows for greater scalability and security. Users of the system perform motor movements in sequence to create a unique password that trains an artificial neural network to correctly identify their specific brain patterns. The brain data collected by the EEG is distilled using Independent Component Analysis and subsequently processed by the binary classification neural network to determine if the user will be authenticated. The machine learning algorithms are trained using the PhysioNet EEG Motor Movement/Imagery Dataset and achieve an accuracy of 100% in the task of deciphering correct brain data from invalid data provided by a different subject. The designed system can be deployed in industries such as banking and finance since the security of this algorithm is unparalleled by other forms of biometric encryption.
The link to the full publication is here: Publication Link