A Survey on Intrusion Detection Systems (IDSs) Using Machine Learning and Deep Learning Techniques

Cybersecurity is becoming an increasingly important field of study because of the growing importance of networks in modern life. The most common cyber security measures include anti-virus software, firewalls, and intrusion detection systems (IDSs). Both internal and external threats can be protected by these methods. An IDS is a detection system that keeps tabs on the health of a network’s software and hardware in order to keep that network’s data safe. Analyzing the software or hardware of a network is the primary function of an IDS, a critical cyber security method. Current intrusion detection systems continue to face difficulties in increasing detection accuracy, reducing false alarm rates, and identifying unexpected threats, even after decades of research. Many academics have focused on developing IDSs that use machine learning approaches to address the issues raised above. Automatic and accurate detection of normal and aberrant data can be achieved through machine learning approaches. In addition, because machine learning (ML) techniques are so generalizable, they may uncover previously unknown attacks. Deep learning (DL) is a branch of machine learning (ML) that has grown in popularity as a result of its superior performance. This study offers an IDS taxonomy based on statistical objects as the primary dimension for classifying and summarizing, ML or DL-based IDS approaches. This form of classification structure, we feel, is appropriate for cyber cybersecurity experts. The survey defines the notion of IDSs and their classification. Furthermore, the MLand DL methods that are often employed in intrusion detection systems, measurements, including benchmark datasets are presented.