Comparision of classification algorithms for network intrusion detection

Network security has become very important nowadays in information system. A large part of the communication between people and institutions is realized on computer networks.We may also have confidential information during this network communication. Confidentiality, integrity, availability is crucial to our knowledge. Malicious people can still our information or take advantage of our information systems by exploiting vulnerabilites on the network. Intrusion detection systems have been developed to protect against such attacks on networked systems today's. At this point algorithms used in intrusion detection systems are of great importance. Because these algorithms vary in terms of performance. In this research, 4 different machine learning algorithms were discovered in WEKA. These learning algorithms are multi layer perceptrons (MLP), support vector machines (SVM), decision tree algorithm (J48) and fuzzy unordered rule induction algorithm (FURIA). In this thesis, these algorithms were compared against each other in terms of performance in the intrusion detection system.