Metigatic Genric Attack for Network Imbalancing Using MLBased CDS Framework

Abstract

Network Intrusion Detection Systems (NIDS) play a vital role in protecting organizations from the growing threat of cyber-attacks that can compromise sensitive data and disrupt critical infrastructure. Traditional IDS face challenges such as imbalanced network traffic and poor feature selection, which reduce detection accuracy. To address these issues, this study proposes a hybrid machine learning-based IDS framework that integrates Conditional Generative Adversarial Networks (CGAN) for balancing datasets and Feature-Guided Population Selection for Particle Swarm Optimization (FIPSO) for optimal feature selection. Transfer learning models like VGG16 and VGG19 were used for feature extraction, and classifiers such as Random Forest, XGBoost, and Decision Tree were employed for attack classification. The hybrid model achieved 96% accuracy, 95% precision, 97% recall, and 96% F1 score in binary classification, outperforming traditional IDS approaches. This research enhances IDS performance by effectively addressing data imbalance and feature optimization, resulting in a more accurate and robust intrusion detection system. newline