Anomaly-Based IDS for Cloud Environments Using Black Widow Feature Selection and Maclaurin–Deep Belief Classifier
Main Article Content
Abstract
Cloud computing is a large domain making efficient use of resources. The key concerns in cloud computing are performance and security. The selection of ideal features and high false alarm rates, required nevertheless for the highest levels of accuracy, are also major challenges. To resolve these problems and improve accuracy, an efficient cloud intrusion detection system (IDS) using the black widow optimization (BWO) algorithm and Maclaurin series–deep belief network (Maclaurin-DBN) is introduced in this paper. BWO is availed to select the ideal features from a feature set. BWO is a new optimization methodology; early convergence and achieving an optimized fitness value are its main advantages. Then a Maclaurin-DBN is deployed to classify the selected feasible features. The BWO’s multiple iterations create the best total neurons, but the hidden layers face different types of attacks. The IDS based on Maclaurin-DBN obtains the maximum identification rate with a compact structure. The introduced deep learning–based IDS is simulated on the Python platform, and the performance of the introduced model is computed and compared with previous deep learning algorithms and existing works. The UNSW_NB15, NSL_KDD, and CICIDS-2017 are the three datasets utilized for the implementation. The simulation outcomes show that the proposed scheme achieved a maximum classification accuracy and detection rate of 95.38% and 93.34%, 99.41% and 98.83%, and 99.42% and 97.12%, respectively, on the NSL-KDD, UNSW-NB15, and CICIDS-2017 datasets.