Optimized CNN-BiGRU Intrusion Detection Model with SMOTE Enhancement: Using Optuna for Automated Hyperparameter Tuning

The primary challenge in the domain of network security revolves around the development of an efficient model capable of detecting intrusions within network systems. Network Intrusion Detection Systems (NIDS) play a crucial role in safeguarding network traffic and preempting potential attacks. NIDS are designed to discern signatures and rule violations from databases, issuing alerts when suspicious activity is detected. Regular updates to the database of traffic signatures are imperative for recognizing the latest network threats. Machine learning and deep learning present promising avenues for identifying legitimate network behavior and thwarting intrusion attempts. In our research, we employ the NSLKDD dataset to train and evaluate our hybrid CNN-BiGRU algorithm. Utilizing the Synthetic Minority Over-sampling Technique (SMOTE) tackles the issue of imbalanced data, enhancing the model's ability to discern patterns in minority classes. Additionally, Optuna is employed for fine-tuning the algorithm, optimizing its parameters for the specific requirements of NIDS. The results of our experiments demonstrate that the suggested approach outperforms other classical intrusion detection models, affirming its efficient performance in network intrusion detection. Achieving an accuracy rate of 98.83% on NSLKDD, the proposed model not only bolstered its ability to identify minority attacks but also maintained a low false positive rate.