Hybrid Detection Model Combining an Advanced Q-Learning Network (AQN) and Attention-Enriched Transfer Learning Approach for DDoS Detection

Modern network systems are seriously threatened by distributed denial of service (DDoS) attacks, which call for sophisticated mitigation strategies. The proposed Advanced Q-learning network(AQN) [1] with an Attention-Enriched Transfer Learning (AETL) [2] is designed to integrate MixNet, Recurrent Neural Network (RNN), and MobileNetV3 with the aid of insights from an advanced classifier. The comprehensive methodology includes data collection, advanced normalization using Recursive Feature Elimination (RFE)[3], anomaly detection-driven data cleaning, and Deep Packet Inspection (DPI) [4]. To improve feature representation, temporal dependencies are captured using Long Short-Term Memory (LSTM) [5] networks and attention mechanisms Adding an attention mechanism, Q-Learning and transfer learning to improve previously trained models allows the suggested method to concentrate on important network traffic features that indicate possible DDoS attacks. The current analysis shows that low false rate in accuracy, precision, and detection speed in finding the mitigation strategies. The proposed model is tested on various datasets and in real-world network environments to confirm its scalability and resilience.