摘要
提出了一种新的异常行为检测方法,将SVM算法和KNN算法结合,在对识别样本判别时,当其与最优分类面的距离大于给定阈值时,采用SVM分类算法,否则采用KNN算法,从而减少了SVM算法的错误率。实验结果表明,SVM-KNN算法对异常行为检测的准确率达到95.86%。
An abnormal behavior detection method is proposed by combining SVM algorithm with KNN algorithm.During the preprocessing phase,when the distance from the test samples to the optimal hyper plane is greater than the given threshold,SVM algorithm is applied to classify test samples,otherwise KNN algorithm would be used to reduce the misclassification probability.Results of experiments show this method of SVM-KNN algorithm has gained the accuracy of 95.86% on abnormal behavior detection.
出处
《辽宁科技大学学报》
CAS
2010年第5期449-452,共4页
Journal of University of Science and Technology Liaoning
