Classification of Cyberattack detection in Network Traffic using Machine learning techniques

As technology advances, so does the number of people using the Internet and the variety of devices that may connect to it. There is now more traffic on the Internet, making it more vulnerable to attacks. In order to identify complex and covert attacks, studies have been concentrating on building learning intrusion detection systems. In order to increase classification performance on attack behaviors with small sample sizes and network attack detection, this paper suggests using Adaptive synthetic sampling (ADASYN). The proposed study uses three machine-learning models for classifying intrusions and attacks on a network. After developing a number of machine learning models—including the K-nearest neighbor, Decision tree, and Random forest—we used Elastic net regularisation-based feature selection to clean, normalize, and oversample for an uneven distribution of labels and shrink the data set. We have tested the proposed models on three publicly accessible benchmark datasets: NSL-KDD, UNSW-NB15, and CICIDS2017. In terms of accuracy and F1-score, the trials favor the hybrid ADASYN with the Random Forest model.