Constructing the lie detection system with fuzzy reasoning approach

Current approaches to lie detection generally rely on specialized instrumentation or environmental conditions which can be time- and cost-intensive to secure and produce questionable results. This study uses electroencephalographic (EEG) variability and fuzzy theory to develop a lie detection model and rule set, identifying sensitive and useful EEG frequency bands to accurately measure lying states based on spectral analysis. Twenty subjects participated in card tests accompanied by EEG recording to evaluate the performance of the proposed model against other data mining methods. The result shows that our proposed model has a lie detection accuracy rate of 89.5% and compares well with other data mining methods. A mobile prototype for real-time lie detection is developed by integrating commercial brainwave measurement instruments with mobile devices. The proposed device can facilitate real-time and accurate polygraphy, while reducing the disadvantages of conventional lie detection approaches.