A data privacy-preserving model driven on differential approach in cloud environment
Rahul Gupta, Deepika Saxena, Ashutosh Kumar Singh · IET conference proceedings. · 2023
With the wide range of cloud services, including flexibility, high scalability, and the least upfront capital investment, it also offers machine learning services that are utilized in various applications, such as medical diagnosis, risk analysis, and product recommendation. Since it has many advantages, many organizations move their data to the cloud platform for storage, processing, and distribution intentions. The classification service is offered by the cloud service provider so that owners can assess their data. It gathers the data from the owners' side for training the machine learning model and accomplishing the classification task. The cloud servers handled by a third party might not be totally trusted by the data owners. However, data privacy is a significant challenge among the critical barriers to preserving data while distributing it. A novel data protection model is proposed, driven on machine learning and differential approaches that carry out classification tasks while protecting the owners' data through noise addition. The model enables the different owners to store, share, and utilize data on the cloud platform. The experiments are carried out on Cleveland, New-Thyroid, and SA-Heart datasets to determine the effectiveness of the model in terms of accuracy, precision, recall, and F1-score. The results show that it achieves high accuracy, precision, recall, and F1-score up to 88.37%, 83.28%, 98.41%, and 82.91%, respectively.