GOA-IDBN-SC: An Improvised Feature Extraction and Classification Approach for Cloud Intrusion Detection

Abstract

Generally, based on the intrusion detection algorithm the user’s information and their behaviors are analyzed, which is the main achievement of the intrusion detection system. Whether the user activity is legal or not and also the essential action over the illegal action is determined here, based on filtering. To detect the known attacks is the foremost problem of existing IDSs, yet new types of unknown attacks were not recognized. In this paper, an efficient Cloud Intrusion Detection system (CIDS) is proposed in which Grasshopper Optimization Algorithm (GOA) is utilized for feature extraction and clustering. An improvised deep belief network with Softmax classification (GOA-IDBN-SC) is devised for classification based on their chosen feasible features, with the deep belief network used for unsupervised feature learning. The IDS is developed on the Python platform and Precision, detection accuracy, f-measure, recall are statistical measurements used to access and equate the performance of proposed model GOA-IDBN-SC with existing deep learning algorithms such as Deep Auto-encoder (DAE), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) and Deep Neural Network (DNN) algorithms.