Privacy-Preserving using Machine Learning based Enhanced Ensemble Algorithm in Cloud Environment

Authors

  • Revati Raman Dewangan
  • Sunita Soni
  • Rakesh Acharya

Keywords:

Artificial Intelligence, Cloud environment, Enhanced ensemble method algorithm (EEMA), Machine learning, Privacy preservation

Abstract

Machine learning algorithms because of profound neural network organizations (NN) have accomplished striking outcomes and are generally utilized in different fields. Then again, as cloud administrations keep on growing. Several machines learning as a service (MLaaS) options are available, where the building and transmission of AI models are done based on cloud service providers. However, AI computations anticipate access to raw data, which is frequently security-sensitive and might provide significant security and protection risks. An ML calculation performs extraordinarily just when taking care of colossal and ideal information for preparation. To get such quality information, numerous associations cooperate agreeably. At the point when we acknowledge information from various associations, it is vital to keep up with the classification, protection and benefit sharing of the information. Homomorphic encryption (HE) is one of the potential cryptographic choices for tackling privacy challenges in machine learning on sensitive data, such as health data and financial data. The huge number of iterations in optimisation techniques like gradient descent (GD) for the learning phase, however, sometimes results in HE-based solutions having relatively high computing costs. It is referred to as ensemble GD. The performance of the enhanced ensemble method based on HE is significantly improved by our ensemble technique since it requires fewer GD iterations and uses less memory.

Published

2023-05-29

Issue

Section

Articles