Deep Learning based Optimization algorithm for Cyber Security Intrusion Detection System
Manasi Gali · Journal of Networking and Communication Systems (JNACS) · 2021
In several networks, intrusion detection plays an important role in assuring cyber security.Numerous studies deal with various cyber attacks in the data via modeling several supervised techniques; however, they have not considered the database size at the time of the optimization.As the size of data increases exponentially, it is vital to cluster the database ahead of detecting the intruder presence in the system.To overcome these confronts, and therefore this paper developed Enhanced Gravitational Search Algorithm -Adaptive Particle Swarm Optimization Algorithm (EGSA-APSO) optimization technique.With the optimization algorithm, the database is clustered into various groups by the developed Intrusion Detection System (IDS) as well as it detects the intrusion presence in the clusters with the employ of the Hyperbolic Secant-based Decision Tree (HSDT) classifier.Subsequently, to the Deep Neural Network (DNN), the compacted data is subjected and train with the optimization method to identify the intrusion detection in the whole database.The experimentation of the developed optimization technique is performed by exploiting several measures such as True Positive Rate (TPR), accuracy, and True Negative Rate (TNR), the outcomes exhibit a superior performance over the conventional models.