计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (20): 158-164.DOI: 10.3778/j.issn.1002-8331.1908-0133

• 模式识别与人工智能 • 上一篇    下一篇

心理测试中掩饰行为的识别研究

赵童,黄钲,王秀超,李淼,张昀,郑秀娟,刘凯   

  1. 1.四川大学 电气工程学院,成都 610065
    2.中国科学院 沈阳自动化研究所 机器人学国家重点实验室,沈阳 110016
    3.中国科学院大学 计算机与控制学院,北京 100049
    4.中国人民解放军空军军医大学 军事医学心理学系,西安 710032
    5.西安交通大学 电子与信息工程学院,西安 710049
  • 出版日期:2020-10-15 发布日期:2020-10-13

Recognition Research on Concealing Behavior in Psychological Test

ZHAO Tong, HUANG Zheng, WANG Xiuchao, LI Miao, ZHANG Yun, ZHENG Xiujuan, LIU Kai   

  1. 1.College of Electrical Engineering, Sichuan University, Chengdu 610065, China
    2.State Key Laboratory of Robotics, Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang 110016, China
    3.School of Computer and Control Engineering, University of Chinese Academy of Sciences, Beijing 100049, China
    4.Department of Military Medical Psychology, Air Force Medical University, Xi’an 710032, China
    5.School of Electronic and Information Engineering, Xi’an Jiaotong University, Xi’an 710049, China
  • Online:2020-10-15 Published:2020-10-13

摘要:

脑电(Electrocorticogram,EEG)信号能够正确地揭示人的心理活动,因此被广泛地运用到心理测试中。提出了一种基于EEG信号的非线性特征融合方法,对受试者在心理测试中是否存在掩饰行为进行识别。对心理测试过程中受试者的EEG信号进行预处理,提取各通道信号的Lempel-Ziv复杂度LZC、样本熵SE、排列熵PE和模糊熵FE四种非线性特征;使用多维尺度分析(MDS)对所得的四种特征的不同特征组合进行融合和降维操作。针对不同特征组合,采用正则化核函数极限学习机构建分类模型并通过测试集验证分类模型的性能。实验结果表明,分类模型准确率能达到82.9%,证明了该方法的适用性。

关键词: 掩饰行为, 非线性特征, 多维尺度分析(MDS)降维, 正则化核函数极限学习机

Abstract:

Electrocorticogram(EEG) can reveal people’s psychological activities correctly, so EEG is widely used in psychological test. In this paper, a method based on the nonlinear feature fusion of EEG signals for concealing behavior recognition in the psychological test is presented. First, after pre-processing the EEG signals of the subject obtained in the psychological test, four nonlinear features, including Lempel-Ziv Complexity(LZC), Sample Entropy(SE), Permutation Entropy(PE) and Fuzzy Entropy(FE), are extracted. Then, the fusion and dimensionality reduction of the features by Multidimensional Scaling(MDS) is done. Finally, the Regularization Kernel Extreme Learning Machine(RKELM) is applied to construct the classifier and the performance of the four trained classifiers is verified by the test set. The experimental results show that the accuracy of the proposed system is 82.9%, which indicate the applicability of the proposed method.

Key words: concealing behavior, nonlinear feature, dimensionality reduction of multidimensional scaling, regularization kernel extreme learning machine