Mechanism and Recent Trends of Federated Learning as a Privacy Technology

Abstract

Recently, constructing and operating platforms have achieved secure data sharing and/or usage between organizations (i.e., cloud services and blockchain networks). According to the latest trends, studies are under way in various fields to improve internal operations and generate new business values using the above platforms. In particular, organizations that use machine learning in their internal processes and/or services focus on federated learning. Federated learning is a technology that can train machine learning models using the datasets stored at different locations without sharing these datasets. Recently, the usage of this technology to improve the efficiency/quality of marketing and internal operation has been increasingly studied. In this paper, we introduce the concept of federated learning and show the recent trends in attack models and their countermeasures and implementation platforms.