Enhancing anomaly detection in cloud computing through metaheuristics feature selection with ensemble learning approach
C. Jansi Sophia Mary, K. Mahalakshmi · China Communications · 2025
Cloud computing (CC) provides infrastructure, storage services, and applications to the users that should be secured by some procedures or policies. Security in the cloud environment becomes essential to safeguard infrastructure and user information from unauthorized access by implementing timely intrusion detection systems (IDS). Ensemble learning harnesses the collective power of multiple machine learning (ML) methods with feature selection (FS) process aids to progress the sturdiness and overall precision of intrusion detection. Therefore, this article presents a meta-heuristic feature selection by ensemble learning-based anomaly detection (MFS-ELAD) algorithm for the CC platforms. To realize this objective, the proposed approach utilizes a min-max standardization technique. Then, higher dimensionality features are decreased by Prairie Dogs Optimizer (PDO) algorithm. For the recognition procedure, the MFS-ELAD method emulates a group of 3 DL techniques such as sparse auto-encoder (SAE), stacked long short-term memory (SLSTM), and Elman neural network (ENN) algorithms. Eventually, the parameter fine-tuning of the DL algorithms occurs utilizing the sand cat swarm optimizer (SCSO) approach that helps in improving the recognition outcomes. The simulation examination of MFS-ELAD system on the CSE-CIC-IDS2018 dataset exhibits its promising performance across another method using a maximal precision of 99.71%.