【摘要】：針對現(xiàn)有測謊方法識別率低的缺陷,將人工免疫算法和超限學習機相結合,提出了一種基于AIA-ELM的N400誘發(fā)電位測謊新方法。將24名被試分成犯罪組和對照組,提取多通道的N400峰值、平均幅值、中值頻率作為特征向量。采用AIA-ELM算法對被試的探測刺激與無關刺激進行分類,犯罪組被試的識別率為97.60%。實驗結果表明,本方法能較有效地進行謊言區(qū)分,為N400測謊提供了一種新的參考依據(jù)。
[Abstract]:Aiming at the defect of low recognition rate of existing lie detection methods, a new method based on AIA-ELM for N400 evoked potential lie-detection is proposed by combining artificial immune algorithm with over-limit learning machine. 24 subjects were divided into crime group and control group. The peak value of N400, mean amplitude and median frequency of multi-channel were extracted as characteristic vectors. The AIA-ELM algorithm was used to classify the detection stimulus and the unrelated stimulus. The recognition rate of the crime group was 97.60. The experimental results show that this method can effectively distinguish lies and provide a new reference for N400 lie detection.
【作者單位】： 陜西師范大學計算機科學學院;現(xiàn)代教學技術教育部重點實驗室;
【基金】：國家自然科學基金(61672021) 陜西省自然科學基金(2017JM6108)
【分類號】：R318.04;TP18

【相似文獻】

相關期刊論文 前1條

1 高軍峰;張文佳;楊勇;胡佳佳;陶春毅;官金安;;基于P300和極限學習機的腦電測謊研究[J];電子科技大學學報;2014年02期

，

本文編號：2397606