Optimization Accuracy of Distributed Denial of Service Attacks in Cybersecurity based on Machine Learning

Abstract

Distributed Denial of Service (DDoS) attacks remain one of the most disruptive threats in modern cybersecurity, causing severe service outages and financial losses across networked systems. This study presents an optimized machine learning-based framework for the accurate detection and classification of DDoS attacks using four widely adopted algorithms: Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), and Extreme Gradient Boosting (XGBoost). The proposed approach emphasizes improving detection accuracy through effective data preprocessing, feature selection, and model tuning techniques. A benchmark network traffic dataset is utilized, where features are normalized and transformed to enhance model performance. Comparative analysis is conducted using key evaluation metrics, including training accuracy, testing accuracy, precision, and recall, to ensure robust and unbiased assessment. Experimental results demonstrate that ensemble-based models, particularly Random Forest and XGBoost, significantly outperform traditional methods such as Logistic Regression and Decision Tree in terms of accuracy and generalization capability. Among all models, XGBoost achieves the highest detection accuracy with improved precision and recall, indicating its effectiveness in handling complex and imbalanced traffic patterns. The findings highlight the importance of optimized machine learning techniques in strengthening intrusion detection systems and mitigating DDoS threats in real-time environments. This research contributes to the development of intelligent, scalable, and efficient cybersecurity solutions capable of enhancing network resilience against evolving attack vectors.