Towards Intelligent Intrusion Detection: Botnet Attack Classification Using ML and DNN

The exponential growth of IoT devices has significantly increased the risk of botnet attacks, whichexploit vulnerabilities in networked systems to execute malicious activities. Traditional botnetdetection approaches often fail to adapt to the dynamic nature of these attacks. The proposedapproach enhances detection capabilities, making it more suitable for large-scale IoT networks. Thisproject proposes a Machine Learning (ML) and Deep Neural Network (DNN)-based approach todetect and classify botnet attacks. Traditional Intrusion Detection Systems (IDS) rely on predefinedrules and struggle with modern threats. To address this, the system uses the UNSW-NB15 dataset,which contains real-world network traffic patterns. Various ML models, including Random Forest,Decision Tree, Naive Bayes, SVM, KNN, XGBoost,,Extra Trees Classifier and a DNN, areevaluated based on accuracy, precision, recall, F1-score, and execution time.The results show thatXGBoost performs best with 94.02% accuracy, followed by Random Forest (93.29%), Extra TreeCl assifier (93.14%), Decision Tree (92.44%), and DNN (90.04%). These models excel in classifyingfive categories: Normal, Exploits, Fuzzers, Reconnaissance, and Generic attacks. This projecthighlights the potential of AI-driven solutions to provide scalable, accurate, and automated botnetattack detection, contributing to stronger and smarter network defense systems.