Subject based Deceit Identification using Empirical Mode Decomposition

Abstract

EEG based lie detectors are gaining attention these days as polygraphs test for deception detection are controlled by human. These detectors use ERP components of EEG to analyze the concealed behavior of any subject. Here in this paper, Empirical Mode Decomposition (EMD) for EEG feature extraction has been applied, as it provide information of both time and frequency domain of a signal. Further, various classifiers have been applied on EEG data to identify subject as guilty or innocent. Classifiers such as SVM, QDA, KNN and decision tree are applied on subject wise EEG data. A novel set of experiments are performed with various participants and analyzed whether they are lying or telling truth. For experimental analysis subjects are presented certain images as stimuli and their responses are recorded and further analyzed. According to results obtained, for most of the subjects, SVM has performed better than others on recorded EEG dataset.