基于听觉E RP功能脑网络特征和SVM的测谎方法研究

基于图论理论的脑网络分析方法近年来在认知脑科学研究中起到了非常重要的作用，而基于事件相关电位（Event-Related Potentials，ERP）的传统测谎方法一直都专注于对某一特定通道上的脑电信号进行分析，针对传统方法中使用少数通道并不能够全面的反映人在说谎状态下大脑整体认知功能特征的缺点，本文提出了基于脑网络特征的测谎方法，通过听觉刺激诱发事件相关电位ERP，记录脑区多通道脑电信号，通过讨论各导联之间的相位延迟指数来构建脑功能网络，计算7类脑网络特征参数作为判别指标。分析被试在说谎和无辜状态下的网络特征参数，使用支持向量机对实验数据进行分类判断，结果表明：本文提出的方法有较高的判别准确率，优于目前判别方法的平均值，证明了本方法的测谎有效性。

Study on Lie Detection Method Based on Auditory ERP Fu nctionaI Brain Network Characteristic and SVM

Recently,brain network method,which based on grapy theory,has played an important role in cognitive science research.And the traditional lie detection methods,which based on ERP signals,usually focus on the EEG from one channel,this has some shortcomings,that use few channels are not able to reflect the whole cognitive characteristic underly-ing lie condition.In this paper,we proposed a method based on brain network characteristics.We used the auditory stimuli to evoke the ERP signals and it was recorded from different channels.In order to build the functional brain network,we calcu-lated the phase lag index between these channels,and seven network parameters were calculated as the index for lie detec-tion.Those network parameters were compared between guilty and innocent subjects,and support vector machine was used as the classifier to the test date.The result shows that this method has a higher identify accuracy than the average accuracy of existing method,proved the validity of the method.