VirusDetection Based on Rough Set and Bayes C lassifier

Rough set and Bayes classifier were integrated in virus detection. On the basis of the feature reduction algorithm of rough se,t the proposed method takes synthetically into account the influences of the dependency of condition features and decision-making features and those of the dependency among condition features on reduction. W ith the method, the limitation on the independence among attribute variables are relaxed, the potential robust properties ofBayes classifier are utilized and its performances are improved. Experiment results show that the detection rate is 97. 88%, and the accuracy is 97. 16%, superior to the signature-based method, RIPPER(repeated incrementalpruning to produce erroe reduction) and Bayesmethod; the false positive rate is 5. 19%, less than the othermethods.