An Efficient Network Intrusion Detection and Classification System

Intrusion detection in computer networks is of great importance because its effects on the different communication and security domains. The network intrusion a challenge. Moreover, remains challenging task as massive amount data required to train state-of-the-art machine learning models detect threats. Many approaches have already been proposed recently detection. However, they face critical challenges owing continuous increase new threats that current systems do not understand. This paper compares multiple techniques develop system. Optimum features are selected from dataset based correlation between features. Furthermore, we propose an AdaBoost-based approach for these present detailed functionality performance. Unlike most previous studies, which employ KDD99 dataset, used recent comprehensive UNSW-NB 15 anomaly collection packets exchanged hosts. It comprises 49 attributes, including nine types such DoS, Fuzzers, Exploit, Worm, shellcode, reconnaissance, generic, analysis Backdoor. In this study, SVM MLP comparison. Finally, AdaBoost decision tree classifier classify normal activity possible We monitored traffic classified it into either or non-threats. experimental findings showed our method effectively detects forms intrusions achieves accuracy 99.3% UNSW-NB15 dataset. system will be helpful applications research