Networks Intrusion Detection Using Optimized Hybrid Network

Networks with critical data can now be protected using a wide range of strategies. A number of dangers exist that could compromise computer networks' integrity, availability, and confidentiality. To identify network-based attacks, intrusion detection systems (IDS) function as defensive tools. It is in the best interest of the research community to create IDS based on artificial intelligence (AI) techniques. Through the use of an improved hybrid model comprising a recurrent neural network (RNN) and a convolutional neural network (CNN), this study suggests a new technique for intrusion or anomaly detection. In this work, some hyperparameters of the model are optimized. For simulation purposes, the NSL-KDD dataset has been utilized, which has four attack categories. The proposed model's performance is then analysed in terms of parameters such as sensitivity, f1-score, precision, specificity, recall, false negative rate, false positive rate, and accuracy. Then the outcomes of the proposed optimized hybrid work are compared with some existing methods. The outcomes discussed in this work confirm that the proposed optimized hybrid machine learning models outperform the listed existing methods in the literature.