DDoS Attack Classification Using Adaptive Multiphase ensemble with Temporal Reinforcement

In the era of rapid digital change, DDoS attacks present a major challenge to network security by disrupting the availability of critical services. Traditional detection methods often struggle with high false-positive rates and delayed response times due to the evolving nature of attack patterns. This paper proposes AMPETR++ (Adaptive Multi-Phase Ensemble with Temporal Reinforcement), a novel hybrid framework for DDoS attack detection and classification using machine learning techniques. The AMPETR++ model integrates three core phases: Statistical Analysis, Frequency-Based Feature Extraction, and Behavioral Analysis, to extract comprehensive features from the network traffic dataset. The model combines Random Forest, XGBoost, and LSTM neural networks with adaptive weight optimization to improve detection accuracy. Additionally, the Temporal Reinforcement Learning Mechanism filters predictions based on past observations, enhancing the model's performance over time. The proposed framework demonstrates superior performance compared to other models, achieving high accuracy, precision, recall, and F1-score on benchmark datasets. The results highlight the model’s ability to dynamically adapt to varying attack patterns, making it a promising approach for real-time DDoS attack detection and prediction in cybersecurity systems.