A Novel Machine Learning-Based Approach for Identification of Unrealistic Tax Returns by Eeg Signal Processing

Abstract

For many years, the uncertainty of lie-detection systems has been one of the concerns of tax organizations. Clearly, the results of these systems must be generalized by a high value of accuracy to be acceptable by related systems to identify tax fraud. In this paper, a new method based on P300-based component has been proposed for detection of tax fraud. To this end, the test protocol is designed based on Odd-ball paradigm concealed information recognition. This test was done on 40 people and their brain signals were acquired. After prepossessing, the classic features are extracted from each single trial. After that, time–frequency (TF) transformation is applied on the sweeps and TF features are produced thereupon. Then, the best combinational feature vector is selected in order to improve classifier accuracy. Finally, guilty and innocent persons are classified by K-nearest neighbor (KNN) and multilayer perceptron (MLP) classifiers. We found that combination of time–frequency and classic features has better ability to achieve higher amount of accuracy to identify the unrealistic tax returns. The obtained results show that the proposed method can detect deception by the accuracy of 91% which is better than other previously reported methods. This study, for the first time, succeeded in presenting a novel method for identifying unrealistic tax returns through EEG signal processing, which has significantly improved the yield of this study compared to the previous literature.