TASK-INDEPENDENT EEG-BASED AUTHENTICATION

Abstract

This paper introduces a cutting-edge approach to Electroencephalography (EEG) based authentication that transcends traditional task-specific requirements, significantly enhancing user experience and authentication accuracy. By employing a convolutional neural network (CNN) to develop a deep learning model, the study successfully extracts feature vectors from EEG signals without necessitating predefined tasks, offering a more adaptable and user-friendly alternative. The proposed system achieved a notable accuracy rate through experiments, including Single-Task and Multi-Task Feature Extraction methods.  The study model achieved an accuracy rate of 95% in authentication by making enhancements to the Multi-Task methodology. These experimental insights underscore the viability and efficiency of task-independent EEG authentication while maintaining robust security measures.