Representation learning, pattern recognition, privacy protection, and security for brain biometric

#STEM #wie

Abstract: Biometrics are body measurements and calculations related to human characteristics, such as fingerprints and faces. They offer a more convenient and secure way for person identification and authentication over traditional methods using passwords and tokens. Brain biometrics based on EEG signals (i.e. brain waves) that reflect electrical activity in the brain is an emerging branch of the biometrics family. Such biosignals have recently been proposed as promising candidates for biometric recognition due to their unique advantages in robustness against forgery, privacy compliance, aliveness detection, and continuous nature. In this talk, I will introduce recent deep learning-based signal processing, feature extraction, and classification methods in brain biometrics, as well as the classical approaches. I will also present our work on representation learning and pattern recognition for brain biometrics and biological signal classification. I will also discuss some open research questions in multimodality, security, permanence, and stability. The second part of the talk will focus on the privacy and security issues of brain biometrics. I will discuss our research on security and privacy protection mechanisms.