南京理工大学学报（自然科学版）2019，Vol.43Issue(5)：586-591,6.DOI:10.14177/j.cnki.32-1397n.2019.43.05.007
一种用于异常检测的实值否定选择算法
Improved real-value negative selection algorithm for anomaly detection
摘要
Abstract
To solve the problem that existing real-valued negative selection algorithms generate redundancy detectors, an improved detector generation algorithm is proposed by analyzing the updating process of the detector set. The proposed algorithm uses dual negative selection and quickly updates the detector set by changing the conditions of accepting and rejecting the null hypothesis, thereby it reduces the generation of invalid detectors. The 2DSyntheticData and the actual Iris data set are used to test the algorithm. Experimental results show that the false positive rate of the algorithm is lower,especially the number of detectors is significantly reduced,and it is suitable for the real-time anomaly detection.关键词
否定选择算法/异常检测/零假设/检测器生成Key words
negative selection algorithm/anomaly detection/null hypothesis/detector generation分类
信息技术与安全科学引用本文复制引用
李亚伦,柴争义,陈国强..一种用于异常检测的实值否定选择算法[J].南京理工大学学报（自然科学版）,2019,43(5):586-591,6.基金项目
国家自然科学基金(61972456) （61972456）
河南省高等学校重点科研项目(14A520079) （14A520079）
河南省科技攻关计划(162102210168) （162102210168）
