Identity Authentication Based on Music-Induced Autobiographical Memory EEG

Abstract

The need for identity authentication has become essential in various aspects of people’s life. In this paper, we propose a novel biometric authentication strategy based on music-induced autobiographical memory electroencephalogram (EEG). Specific music is used to induce the stable autobiographical memory, while the EEG signals are collected through the memory process. Users can authenticate themselves by recollecting their minds when listening to the music, which is closely related to their long-term memory. Based on six types of EEG features from 12 subjects, mean F1 score of 0.937, 0.936 and 0.968 are achieved using Logistic Regression, Support Vector Machine and RUSBoost classifier, respectively. This promising result indicates the high distinctive characteristics in music-induced autobiographical memory EEG, which is suitable for identity authentication applications.