A Relative Vision of Privacy Preservation Over Machine Learning Proposal

Abstract

Several domins have been used such machine learning design on Deep Neutral Network (NN) to accomplish most important activities. The rise of cloud computing implies that many machine learning as services are provided (MLaaS) that include preparation and the delivery of machine learning models in the networks of cloud providers. Machine learning algos, therefore, need exposure to the raw data that is frequently vulnerable to privacy and can pose possible privacy risks. To cope with this issue, we compare different machine learning approaches like feedforward, backpropagation, and SVM in terms of the preservation of privacy. As a standard ML model, SVM makes effective data grading and finds use in real-life situations such as disease diagnosis or anomaly detection. Comparative protection research has demonstrated that the protected SVM system guarantees that confidential details are held secure and healthy. Considerable studies denotes that the comparative scheme is successful.