Research of anomaly detection algorithm based on cloud theory
Dong Wang · Jisuanji yingyong yanjiu · 2009
In order to solve the problems that detection rate was low and the false positive rate was high,this paper proposed an anomaly detection algorithm based on cloud theory.The algorithm abstracted features and computed impact factors of individual attribute by applying chi-square method,introducing cloud generator to compute the feature value and error of attributes,judging intrusions according to integrated-value,which reduced the partial effect due to attribute abnormal overly.Taken the algorithm by tested though KDD'99 data set,the result of simulations demonstrates that excellent performance of the proposed algorithm's average detection rate is 98.66%,while the false positive rate is 1.87%.In a certain extent,the algorithm solves the problems in intrusion detection algorithms.