Metaheuristic-Enhanced Feature Selection for High-Accuracy Intrusion Detection in Cloud Computing

The demand for novel Intrusion Detection Systems (IDS) has risen due to the increasing volume, velocity, and variety of data collected through modern wireless networks. Metaheuristic algorithms provide a viable method to improve IDS performance regarding optimal feature selection. The integration of methods with Machine Learning (ML) in the development of an Intrusion Detection System (IDS) facilitates improved detection accuracy, diminished false positives and negatives, and increased efficiency in network monitoring. This paper presents a hybridization of Grey Wolf Optimization and Lion Optimization, termed GWO-LOA-RF Optimization Algorithm, integrated with a Machine Learning Assisted Intrusion Detection System. The primary objective of the GWO-LOA-RF system is the effective identification and categorization of intrusions to ensure security. Data normalization is primarily conducted to scale the incoming data into a functional format. The GWO-LOA-RF approach features the GWO-LOA-RF base. The random feature selection method can be utilized to identify an optimal subset of features. We have derived multiple insights on feature selection scores in relation to accuracy, recall, precision and F-score as well as Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error for the two datasets, CICDDOS2019 and UNSW-NB15. The duration of feature selection has been emphasized to illustrate the most efficient algorithm combinations. We have ultimately presented hybrid algorithmic to organizational IDS requirements, along with distinct solutions for each.