MLPAM: A Machine Learning and Probabilistic Analysis Based Model for Preserving Security and Privacy in Cloud Environment
Ishu Gupta, Rishabh Gupta, Ashutosh Kumar Singh, Rajkumar Buyya
Abstract
The organizational valuable data needs to be shared with multiple parties and stakeholders in a cloud environment for storage, analysis, and data utilization. However, to ensure the security, preserve privacy while sharing the data effectively among various parties have become formidable challenges. In this article, by utilizing encryption, machine learning, and probabilistic approaches, we propose a novel model that supports multiple participants to securely share their data for distinct purposes. The model defines the access policy and communication protocol among the involved multiple untrusted parties to process the owners' data. The proposed model minimizes the risk associated with the leakage by providing a robust mechanism for prevention coupled with detection. The experimental results demonstrate the efficiency of the proposed model for different classifiers over various datasets. The proposed model ensures high accuracy and precision up to 97% and 100% relatively and secures a significant improvement up to 0.01%, 103%, 151%, 87%, 96%, 43%, and 186% for average probability, average success rate, detection rate, accuracy, precision, recall, and specificity, respectively, compared to the prior works that prove its effectiveness.