Conditional Generative Adversarial Network with Optimal Machine Learning Based Intrusion Detection System

Intrusion Detection System (IDS) is intended to analyse and monitoring system actions or network traffic to recognize and relate to capable security attacks or intrusions. If managing imbalanced data as part of IDS, it suggests that the classes (normal vs. intrusions) could not be defined equally from the database. This condition is usual in security-related databases; as normal network traffic is often significantly greater than malicious traffic (intrusions). Addressing the challenges of imbalanced data has been vital to creating effectual IDS. If not managed properly, the IDS might be biased nearby the majority class (normal traffic) and may miss detecting extremely harmful intrusions. Therefore, this study presents a Conditional Generative Adversarial Network with Optimal Machine Learning based Intrusion Detection System (CGAN-OMLIDS) technique. The goal of the CGAN-OMLIDS technique is to address the class imbalance data problem and detects intrusions. To obtain this, the presented CGAN-OMLIDS technique initially applies the Z-score normalization process. In addition, the CGAN-OMLIDS technique utilizes CGAN method for handling class imbalance problems. Moreover, the Extreme Learning Machine (ELM) approach is exploited for the classifying and detecting the intrusions. Besides, the parameter tuning process of the ELM approach is achieved by the Lion Optimization Algorithm (LOA). The investigational outputs of the CGAN-OMLIDS approach are examined on benchmark databases. The stimulation values highlighted the enhancement of the CGAN-OMLIDS approach over other current techniques.