Lie Detection Using Fuzzy Ensemble Approach With Novel Defuzzification Method for Classification of EEG Signals

The ability to analyze deceit behavior in humans is an extremely vital issue where legal and security purposes are considered. Using the P300 component of event-related potential (ERP) [electroencephalography (EEG)], we have developed a concealed information test (CIT), which provides an easy and effective way to identify whether an individual carries some concealed information. The P300 component is elicited when an individual is provoked with stimuli related to his/her memories. Taking advantage of this P300 feature, several images of certain familiar and nonfamiliar faces are shown to the subjects during the experiment. Throughout the image presentation, EEG data of the subject are recorded. Signal preprocessing, processing, and classification algorithms are then implemented on the recorded EEG data. This article aims to classify the recorded EEG data for CIT. For extracting the spatial features from the recorded EEG data (of range 0.5-30 Hz), a common spatial pattern has been used. A fuzzy integrator system is developed using classifier performance measures as the antecedents. A novel defuzzification function called generalized mean of maxima (GMoM) has also been proposed to achieve a classification score or value. The experimental results demonstrate the outperformance of fuzzy-based CIT with significant classification accuracy.