Network intrusion detection using feature selection and PROAFTN Classification

Classification and detection of network based intrusion is very critical task. The processing of classification and detection faced a problem of large number of attribute and mixed category of data. Day to day increases diversity of attacker and hacker generates new pattern of file for attack purpose, the process of classification and detection suffered due to this reason. The process of classification of PROAFTN is a combination of fuzzy logic and protein cell classification technique. In PROAFTN classification process all features come to the predefined classes for the classification. Now the process of improvement need some important feature selection process for increasing the classification ratio and process of classification. The particle of swarm optimization is dynamic population based searching technique. In the searching technique of particle of swarm optimization select optimal feature set for the classification process of PROAFTN classification process. The process of optimal selection of feature set increase the classification and detection ratio of modified PROAFTN classification process. For the evaluation of performance of modified PROAFTN classification technique used MATLAB 7.8.0 software. MATLAB is well known algorithm analysis software. For the analysis of PROAFTN classification process used fuzzy set tools and some standard tools of MATLAB. For the processing of input data used KDDCUP99 dataset. KDDCUP99 dataset is well known dataset for the purpose of network based intrusion detection and classification. Our classification and detection ratio in some attack case achieved 100%.