Intrusion Detection System Using Deep Belief Network & Particle Swarm Optimization

Securing the services of security such as data integrity, confidentiality and availability is one of the great challenges. Failure to secure above will potentially lead many cyber-attacks. One of the greatest hits for detecting intrusion is an intrusion detection system (IDS) and there are so many advances put forward by many researchers. Even though there exists a large number of Intrusion Detection Systems intruders are still continuing with their job. Another evolving and yet revolutionized strategies is Deep Learning. So, integrating these two systems to create an effective model that could potentially find normal or malicious attacks. In this paper, we classify intrusion using Deep Belief Network and Particle Swarm Optimization into categories like Normal, Probe, DoS, U2R, R2L. The dataset used for applying this model is DARPA 1999 and they are evaluated under various measures. Also, the proposed system is compared with other system like ANFIS, HHO, Fuzzy GNP in which our system outperforms better with greater accuracy of 96.5%.