AI-Driven Real-Time Network Intrusion Detection using Machine Learning and Deep Learning
Vivek Bagmar, Divyanshu Joshi · 2025
With the ubiquity of digital connectivity, cyberattacks on network infrastructures have become more sophisticated and abundant. This paper suggests an artificial intelligence (AI)-based real-time network intrusion detection system (IDS) by utilizing machine learning (ML) and deep learning (DL) models. Using a derived binary classification model from real-time heuristics, the research focuses on two significant features, srv_serror_rate and rerror_rate, to label network traffic as either "Anomalous" or "Normal." Comparative study of Neural Networks, K-Nearest Neighbors (KNN), Random Forests, and Naive Bayes models are illustrated. The models are compared on the basis of metrics like Accuracy, Precision, Recall, and F1 Score. Detailed visual analysis by way of radar plots, confusion matrices, and ROC plots gives model performance, stability, and deployability in real network environments. Among the tested models, the Random Forest algorithm performed the best with an accuracy of 99.80%, precision of 99.69%, recall of 99.66%, and F1 Score of 99.67%, which makes it suitable for use in real time. The results affirm the excellent performance of the Random Forest model and suggest its appropriateness for use in real-time IDS.