Utilizing ML and DL Algorithms for Alert Classification in Intrusion Detection and Prevention Systems: A Detailed Review
Utkarsh Dixit, Suman Bhatia, Pramod Bhatia · 2022 2nd International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE) · 2022
Intrusion detection/prevention systems have attracted much interest in recent years due to increased online connectivity. In recent years due to COVID pandemic and due to the increased number of online users, online data has become more and more exposed to different types of attacks. Hence, in order to keep data safe, it has become quite important to detect/prevent such attacks. An IDS is a sensor that is used for the observation of such attacks on the nodes or the network itself, and in this way, it tries to keep the information safe from possible attacks. However, accurately identifying such attacks so that they can be prevented effectively is a concern. This accuracy is measured by the number of false positive & false negative in a dataset. These days ML/DL algorithms are being significantly utilized for improving the accuracy of different systems (e.g., health care, stock market, forecasting etc.). Considering its importance, the work presented here studies the impact of using ML/DL algorithms on the accuracy of IDS/IPS. The impact of these algorithms is studied by using evaluation metrics for classification of network assaults in the intrusion detection system using different datasets. These algorithms are subject to further changes for improving the accuracy parameters based on evaluation metrics.