Deep Q-Learning Based Reinforcement Learning Approach for Network Intrusion Detection

The rise of the new generation cyber threats demands more sophisticated and intelligent defense solutions equipped with autonomous agents capable learning to make decisions without knowledge human experts. Several reinforcement methods (e.g., Markov) for automated network intrusion tasks have been proposed in recent years. In this paper, we introduce a detection method, which combines Q-learning based deep feed forward neural method detection. Our Deep Q-Learning (DQL) model provides an ongoing auto-learning capability environment that can detect different types intrusions using trial-error approach continuously enhance its capabilities. We provide details fine-tuning hyperparameters involved DQL effective self-learning. According our extensive experimental results on NSL-KDD dataset, confirm lower discount factor, is set as 0.001 under 250 episodes training, yields best performance results. also show highly detecting classes outperforms other similar machine approaches.