Next-Generation Intrusion Detection Systems with Machine Learning and Deep Learning-Based Firewalls
Riturani · 2025
The growing complexity of cyber-attacks has underlined the requirement of strong Intrusion Detection Systems (IDS) able to instantly recognize a variety of hazards. This work offers a strategy for improving Next-Generation Intrusion Detection Systems (NGIDs) combined with firewalls by use of machine learning (ML) and deep learning (DL) approaches. Using the NSL-KDD dataset—which includes both normal and attack network traffic—the work creates a system able to classify traffic patterns into "Normal" and "Attack" categories. The proposed method consists in four phases: data collecting, data preprocessing, feature selection and extraction, and classifier application. In the preprocessing stage, the dataset is cleaned and normalized; hence, suitable feature selection follows to increase classification accuracy. Trained and tested for classification using performance criteria including accuracy, precision, recall, and F1-score, naïve bayes and support vector machine (SVM) models are then compared in efficacy. The results reveal that the SVM classifier has low misclassification and great accuracy (97.29%), therefore proving its prospective use in advanced IDS. The article underlines how several models could be merged inside a firewall system to offer real-time network protection. Future studies will look at using deep learning techniques and ensemble methods to raise detection accuracy even more, therefore allowing NGIDS to be more adaptable and efficient in countering evolving cyber threats.