Classifying DDoS Attacks Using Machine Learning

Authors- Syed Nabiel Basha K, Associate professor Dr S R Raja

Abstract-Distributed Denial-of-Service (DDoS) attacks pose a significant threat to the stability and security of network infrastructures, resulting in severe disruptions and economic damage. This work uses machine learning approaches to classify and effectively mitigate DDoS attacks. Utilizing a labeled dataset of network traffic, important attributes such as packet flow rates, source entropy, and protocol distribution are investigated for training and testing several machine learning models. Paper on Classifier Performance Comparison: Decision Trees, Random Forests, and Neural Network Detecting Normal vs. Malicious Traffic. Results The results indicate the best model was [Random Forest] at a very high accuracy of 100%. Real-time attack detection highly depends on feature engineering with hyperparameter tuning. This scalable and efficient research contributes toward network resilience by ensuring early detection of DDoS. Future work may involve the dynamic retraining of the model to capture changing attack vectors and will also aim to integrate this solution with intrusion prevention systems.

DOI: /10.61463/ijset.vol.12.issue6.374