An Efficient Meta-Heuristic Dimensionality LSTM Model for Dimensionality Reduction and Clustering for Attack Classification

Intrusion detection helps secure and monitor systems and networks for malicious activity. As technology has advanced, network systems have become more vulnerable to intrusion. More than simple firewalls and detection systems-traditional ones are required to defend against advanced persistent threats. When breaches occur, they must be addressed carefully. Modern Security Information and Event Management (SIEM), a current firewall, and an intelligent Intrusion Detection System (IDS) must be used to protect an organization's infrastructure. This paper proposed a Gaussian Meta-Heuristics Attack Classification (GMAC) for attack detection and classification. The GMAC model uses the meta-heuristics model for the clustering process. The Gaussian Mixture Model (GMM) in the meta-heuristics model feature is optimized, and dimensionality reduction is performed. The GMAC model uses the Long Short-Term Memory (LSTM) model for the deep learning model for attack detection and classification. The GMAC model's efficiency is evaluated when considering attack platforms such as KDD‘99 and NSL-KDD. The performance analysis stated that the GMAC model achieves a classification accuracy of 99%.