面向异常检测的双重否定黑洞覆盖算法  

Dual negative algorithm for black hole coverage of anomaly detection

作　　者：胡博[1] Hu Bo(College of Science and Technology,Jiangxi Normal University,Nanchang 330027,China)

机构地区：[1]江西师范大学科学技术学院,江西南昌330027

出　　处：《南京理工大学学报》2018年第5期604-608,共5页Journal of Nanjing University of Science and Technology

基　　金：国家自然科学基金(U1604612)

摘　　要：为了提高实值否定选择算法对黑洞区域判断的科学性和准确性,该文提出一种改进的双重否定黑洞覆盖算法。对于与检测器匹配的数据,所提算法不再将其简单地标识为正常数据,而是通过与自体数据的比较判断,将其标识为正常数据或异常数据。在数据集2DSyntheticData上的实验结果表明,该算法提高了对异常数据的检测率,方差较小,检测结果较稳定。Aiming to improve the scientificalness and accuracy of the existing real-valued negative selection algorithm,an improved dual negative algorithm for the black hole coverage is proposed.For the data not matching the detector,the proposed algorithm does not simply identify it as the normal data,but identifies it as the normal or abnormal data by further judgment.The 2DSyntheticData is used to test the algorithm.Experiment results show that the algorithm can improve the detection rate of the abnormal data,the variance is small and the test results are more stable.

关 键 词：异常检测 人工免疫系统 否定选择算法 黑洞 检测率 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP309[自动化与计算机技术—控制科学与工程]