To search, Click below search items.


All Published Papers Search Service


An Efficient Hybrid Classifier Model for Anomaly Intrusion Detection System


Asghar Ali Shah, M. Khurram Ehsan, Kashif Ishaq, Zakir Ali, Muhammad Shoaib Farooq


Vol. 18  No. 11  pp. 127-135


Ensuring security has always been a challenging problem for both customized network solutions and information systems. Intrusion Detection System (IDS) is playing a very important role to ensure security both in network solutions and information systems. Significant efforts has already been made and many efforts are underway to improve the IDS but still there are many short comings. This study proposed a model based on extensive survey to create an efficient hybrid classifier which is jointly based on feature selection, parameter optimization and classification. Feature selection is adapted to refine the area of interest by improving the accuracy of classification, then to optimize the parameters, genetic algorithm (GA) is the most appropriate technique to be used. Parameters optimization using GA also plays a remarkable role to improve classification using support vector machine (SVM). SVM is considered a suitable machine learning technique for classification of intrusions which are detected both in networks and information systems. Finally SVM will classify the observed activity as normal or attack using adopted linear or nonlinear techniques. The proposed solution paves a way to improve accuracy by efficiently detecting the intrusions within real time applications of network and systems.


IDS, Classification, SVM, GA, Attacks types