A network intrusion detection system based on a Hidden Naïve Bayes multiclass classifier

摘要: With increasing Internet connectivity and traffic volume, recent intrusion incidents have reemphasized the importance of network detection systems for combating increasingly sophisticated attacks. Techniques such as pattern recognition data mining events are often used by to classify either normal or attack events. Our research study claims that Hidden Naive Bayes (HNB) model can be applied problems suffer from dimensionality, highly correlated features high stream volumes. HNB is a relaxes method's conditional independence assumption. experimental results show exhibits superior overall performance in terms accuracy, error rate misclassification cost compared with traditional model, leading extended models Knowledge Discovery Data Mining (KDD) Cup 1999 winner. performed better than other state-of-the art models, SVM, predictive accuracy. The also indicate our significantly improves accuracy detecting denial-of-services (DoS)