Deep Learning-based Intrusion Detection Systems: Enhancing Cyber Defense Mechanisms

Despite the physical and temporal fluctuations in the attributes of cloud environments that are ever-changing, customers are turning to use cloud services more, although it is a menace to users. Most of the intrusion detection systems (IDS) do not have the capacity to accommodate the evolving characteristics of existing threats in cyberspace. This work explores deep learning applications towards the enhancement of intrusion detection system under cloud environment. It is supposed to respond host based and data based security threats. A better accuracy and versatility in real time threat identification (RTTI) can be realized in using the deep learning IDS incorporating the autoencoders, recurrent neural networks (RNNs) and the convolutional neural networks (CNNs). The evaluation of the performance of this system is conducted with the help of experimentation based on benchmark datasets, such as NSL-KDD and CICIDS2017, which portray different types of attacks situations. The evaluation of the model is carried out by precision, accuracy, recall, F1 score, false positive rate and takedown time. Some of the important operational issues which have been highlighted in this paper include the place of IDS in the cloud-based systems and scaling performance and levels of latency in the system. Project results indicate that this was achieved in real-time having superior performance compared to the traditional machine learning algorithms in the threat detection jobs. Deep learning being applied in the design of intrusion detection system is one way through which this designed system can learn better and improve the intelligence level to fit the specific features of the cloud.