SC-CVAR: Intrusion Detection Using Feature Selection and Machine Learning Techniques on UNSW-NB15 Dataset

This research study provides an effective mechanism in detecting and classifying the attacks taken from UNSW-NB15 dataset such as backdoor, Exploits, Shellcode, analysis, fuzzers, generic, normal, reconnaissance, DoS and Worms attacks. Specifically, to enhance the accuracy, the feature selection process is performed using SC-Sine Cosine algorithm, selected only the significant features. Finally, the classification of the intrusion is performed using the Novel CVAR- k fold Cross validated Artificial neural network weighted Random Forest classification. In the prediction phase the type of attacks are revealed. Finally, the proposed SC-CVAR model evaluated in terms of different performance metrics and compared with various existing models to prove its efficiency. The research outcome revealed that this research is highly effective in detecting and classifying the attacks in greater accuracy