PERFORMANCE EVOLUTION OF MACHINE LEARNING ALGORITHMS FOR NETWORK INTRUSION DETECTION SYSTEM

Network Intrusion Detection System (NIDS) is one of the best solutions against network attacks. Attackers also dynamically change tools and technologies. However, implementing an associated accepted NIDS system is an additional challenge. This paper conducts and analyzes many experiments to evaluate numerous machine learning techniques that support the NSL-KDD intrusion data set. We have succeeded in identifying a number of performance metrics to judge the chosen technology. The main focus was on accuracy, precision and recall performance metrics to enhance the detection rate of network intrusion detection systems. Experimental results show that the deep learning approach achieves the highest accuracy and detection rate, while false negatives and false positives are rarely achieved.