Unsupervised anomaly detection algorithm based on D-S evidence theory
In practical applications,when the dataset is unlabeled or the amount of normal point data is insufficient usually leads to one-class support vector machine(OCSVM)operating in an unsupervised manner.In addition,when the training set includes anomalies,the decision boundary formed by OCSVM will skew toward to the anomalies.The above problems may undermine the detection rate of anomalies and result in poor performance of the classifier.In order to solve the above problems,we divided the dataset into suspicious normal point dataset and suspicious outlier dataset based on KNN algorithm.The suspicious normal point dataset was used for OCSVM training and modeling,and for the suspicious outlier dataset,the D-S evidence theory was utilized to identify the normal data of them.The experiment results showed the DS-SVM algorithm can effectively separate the normal points and the anomalies,the mean Auc value of algorithm was 0.83 on the overall dataset,and 0.883 in the suspicious outlier dataset.
outlier detectionone-class support vector machinedempster-shafer evidence theoryunsupervised learning
NETL
NSTL
万方数据
