Improving Intrusion Detection by the Automated Generation of Detection Rules

摘要: Rule Based Detection Systems have been successful in preventing attacks on network resources, but suffer a problem that they are not adaptable cases where new made i.e. need human intervention for investigating attacks. This paper proposes the creation of predictive intrusion detection model is based usage classification techniques such as decision tree, Naive Bayes, neural network, and fuzzy lo gic to generate rules. The proposed this consists two stages. first stage uses either Decision tree (J48 C4.5) or Bayes classifier results obtained experiments while second hybrid module both (MLP) logic. Training evaluation phases used randomly selected connections subset KDD’99 data set. A set features has extracted from those using algorithm. shows how system trained detailing parameters affect training process; it also details process including false positive rates.