Evaluation of Machine Learning Models for Intrusion Detection
P Ishwarya, R Yogeeswar, S. Pavithra · 2025
An IDS is a mechanism of security for monitoring either network traffic or system activities for signs of malicious actions or policy violations. IDS plays a very vital role to identify potential threats to the security of computers and other networks against unauthorized access to them. It has developed as a most powerful means to improve the capability of the IDSs using machine learning, which detects attacks based on automated observations and adjustment to changing patterns of the attacks. The discussion here is with regard to some of the applications of machine learning on these techniques: Random Forest, Support Vector Machine, and Gradient Boosting. These applications are illustrated with the implementation of an IDS over a dataset. It later goes on to evaluate several key performance metrics such as accuracy, ROC curves, confusion matrices, and classification reports. The results on implementation illustrate that the Random Forest model stands out as best, where the highest testing accuracy is shown to be 99.969%, followed closely by SVM at 99.879% and GB at 99.771%. Further evaluation has identified the high classification capability of these models, hence suitable for reliable intrusion detection. This experiment has portrayed that the techniques based on machine learning do improve the precision of the IDSs; therefore, better security can be achieved in terms of preventing unwanted intrusions.