首页>
外国专利>
A detection method based on k-nearest neighbor algorithm for system anomaly
A detection method based on k-nearest neighbor algorithm for system anomaly
展开▼
机译:基于k近邻算法的系统异常检测方法
展开▼
页面导航
摘要
著录项
相似文献
摘要
#$%^&*AU2018100222A420180322.pdf#####Abstract: This invention is a way of detection in system exceptions by using K-nearest neighbor algorithm. The method will define the user system' s instruction set to a converted characteristic, and distinguish the data set to training data and testing data. By using the K-nearest neighbor, it recognizes the data sets and predicts the recognition results to estimate the system exceptions. If there are exceptions, it will alarm the user. The method is easily to approach, high accuracy with strong practicability. 1User1 user_cmnd_feature fdist User2 user cmd feature fdiSt User50 user_CMd_feature Data fdist user1_l abel L user50_label Figure 1 -Etra ] accuaryDifferent K * Pick Kopt accuary Fig.2 New normal o command operation list Use Kopt Yes New User Data Process- History Data KNN a abnormal Warning operation Fig.3 1
展开▼