DATA PRIVACY-PRESERVING VIA IMPROVED FEDERATED LEARNING MODEL
Main Article Content
Abstract
Data modeling is an important problem in data analysis. Machine learning is the most popular method to solve this modeling problem. All most of machine learning schemes are local learning schemes in which the training dataset is stored at a server, therefore it can’t take advantage of the diversity of data shared from multiple sources. As a result, the generalization of the obtained model is limited. The federated learning is a learning from multi-source of data so it has many advantages compared to other methods. Federated learning model applies to a variety of data types and machine learning algorithms. Besides accuracy, this model also ensures privacy for the training data set. This paper proposes an improvement of the federated learning model to ensure privacy protection based on an federated-learning model. The experimental results show the feasibility which can be applied to problems using machine learning in practice and also open up challenges to improve research and innovation.
Keywords
differential privacy, federated learning, privacy-preserving data analysis, privacy-preserving with federated learning
Article Details
References
Fanglin Li, Bin Wu, Liutong Xu, Chuan Shi, & Jing Shi (2014). A fast distributed stochastic Gradient Descent Algorithm for Matrix Factorization, Beijing Key Lab of Intelligent Telecommunication Software and Multimedia.
Jakub Konecny, H. Brendan McMahan, & Daniel Ramage (2016). Federated Optimization: Distributed Machine Learning for On-Device Intelligence, University of Edinburgh.
McMahan, Daniel Ramage (2017). Federated Learning: Collaborative Machine Learning without Centralized Training Data.
Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, Blaise Aguera y Arcas (2017). Communication-efficient learning of deep networks from decentralized data, Google, Inc., 651 N 34th St., Seattle, WA 98103 USA.
Saad Albawi, Tareq Abed Mohammed, & Saad Al-Zawi (2017). Understanding of a convolutional neural network, Department of Computer Engineering, Istanbul Kemerburgaz University, Istanbul, Turkey.
Siddharth Sharma, & Simone Sharma (2020). Activation functions in neural networks, Dept. of Computer Science and Engineering, Global Institute of Technology, Jaipur.