Guilty Knowledge Test using Recurrence Plot Analysis and Channel Information Fusion Technique

In recent years, using the P۳۰۰ component in the guilty knowledge test has become one of the most widely used methods. In various studies, the Pz channel is known as a channel containing the most important information about the P۳۰۰. However, other studies have shown that the Fz and Cz channels also provide useful information about the P۳۰۰ component. Therefore, simultaneous and optimal use of information from these three channels can effectively improve the accuracy of the guilty knowledge test. In this paper, two approaches to classify the guilty and the innocent subjects were used. Also, to extract information related to the P۳۰۰ component, recurrence quantification analysis of ERP single-trials was used. In the first approach, each channel's recurrence plot feature vector was classified independently and applied to identify guilty and innocent subjects using SVM. Then, a decision-level fusion approach was used to fuse the chaotic information of these three channels through the majority voting procedure to determine the final label of each subject. The results of the decision-level fusion approach with ۸۶.۶۷% accuracy, ۸۶.۶۷% sensitivity, and ۸۶.۶۷% specificity show this method's supremacy compared to classification results of each channel independently.

کلیدواژه ها:

Guilty knowledge test ، EEG ، P۳۰۰ component ، Recurrence plot ، Information fusion ، Majority voting procedure