Hybrid Deep Learning Model for Cyber-Attack Detection

Better detection and avoidance of intrusions solutions are in high demand because of the worldwide upsurge in hacking on computer networks. New technologies like fog computing, cloud computing, and the Internet of Things have dramatically increased the potential for cyberattacks and other forms of cyber risk. These attacks can compromise computer network infrastructures, web services, and social media platforms, resulting in economic and reputation loss. Because of its usefulness in detecting and halting malicious actions, intrusion detection systems (IDS) play a crucial part in network defence mechanisms. In this study, a hybrid Deep Learning (DL) network was used to identify the cyberattack. This effort began with the collection and processing of cyber-attack data from the NSL-KDD. Convolutional Neural Networks (CNNs), Bidirectional Long Short-Term Memory (Bi-LS TMs), and hybrid CNN+Bi-LSTMs are the DL models trained with 80% of processed data. The remaining 20% of the data is used for testing the models after they have been trained. Both positive and negative metrics are used to assess the results of the testing phase. When compared to other networks, the recommended CNN+Bi-LSTM model achieves the highest score for positive metrics and the lowest score for negative metrics.