基于P300和极限学习机的脑电测谎研究

发布时间：2018-07-12 12:22

本文选题：脑电 + 极限学习机　；参考：《电子科技大学学报》2014年02期

【摘要】：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。
[Abstract]:The LLM is based on a typical single hidden layer feedforward neural network (SLFNs) and its effectiveness has been verified in many fields of pattern recognition. Aiming at the shortcomings of the current lie detection methods, which are not accurate enough and the training time is long, the ELM algorithm is applied to the field of lie-detection research, which is used as a classifier to classify and recognize the two kinds of EEG signals of liars and honest people. The experimental results are compared with those of three typical classifiers: support vector machine (SVM), artificial neural network (Ann) and linear classifier (FDA). The experimental results show that this method not only achieves the highest training and test accuracy, but also shortens the training time, which proves the validity of the method.
【作者单位】：中南民族大学认知科学国家民委重点实验室;电子科技大学生命科学与技术学院;江西财经大学信息技术学院;
【基金】：国家自然科学基金(81271659,61262034,91120017) 江西省自然科学基金(20114BAB211020,20132BAB201025) 江西省教育厅科技项目(GJJ13302)
【分类号】：R318.0

【共引文献】