Extreme learning machine (ELM) is a typical SLFN (single layer feedback network) and its efficiency has been proved by many literatures for pattern recognitions. In this paper, ELM is applied in lie detection for the first time in order to overcome the disadvantages of the current lie detection methods such as lower accuracy and slower training speed. ELM is used as a classifier to classify the guilty and innocent subjects. The experimental result is compared with support vector machine (SVM), artificial neural network (ANN) and fisher discrimination analysis (FDA). The comparison results show that the proposed method obtains the highest training and testing accuracy with the fastest training speed.%极限学习机基于一种典型的单隐层前馈神经网络(SLFNs)，其有效性在模式识别很多领域得到证实。该文针对当前的测谎方法的准确率不够高及训练时间较长的缺点，将ELM算法应用到测谎研究领域，作为分类器，对说谎者和诚实者的两类脑电信号进行分类识别，并将实验结果和三类典型的分类器：支持向量机(SVM)、人工神经网络(ANN)和线性分类器(FDA)的分类结果进行比较。实验结果表明，该方法不仅获得最高的训练和测试准确率，而且训练时间也大为缩短，证明了该方法的测谎有效性。
展开▼
