Privacy-Preserving Machine Learning Techniques for Data in Multi Cloud Environments

Abstract

Cloud computing has significantly transformed how organizations handle their data. By utilizing multiple cloud services, businesses can achieve greater operational flexibility and cost savings. However, this approach also introduces new challenges, particularly concerning the privacy and security of information. This research explores the application of differential privacy in machine learning, specifically focusing on logistic regression models trained on the MNIST dataset. Key findings include the model’s performance, the privacy-accuracy trade- off, and generalization capabilities. This study demonstrates the feasibility and effectiveness of differential privacy in creating machine learning models that respect data privacy without significantly compromising accuracy.