Naïve Bayes Decision Tree Hybrid Approach for Intrusion Detection System

Abstract

Internet is also increasing exponentially increasing intrusion or attacks by crackers exploit vulnerabilities in Internet protocols, operating systems and software applications. Intrusion or attacks against computer networks, especially the Internet has increased from year to year. Intrusion detection systems into the main stream in the information security. The main purpose of intrusion detection system is a computer system to help deal with the attack. This study presents a hybrid approach to decision tree algorithm and naïve Bayes to detect computer network intrusions. Performance is measured based on the level of accuracy, sensitivity, precision and spesificity. Dataset used in this study is a dataset KDD 99 intrusion detection system. Dataset is composed of two training data and testing data. The selection of attributes is done using the chi-square, selected the top ten attributes based on the calculation of chi-square. From the experimental results obtained by the accuracy of naïve Bayes decision tree algorithm was 99.82%.