Improved Gradient Inversion Attacks and Defenses in Federated Learning

IEEE Transactions on Big Data, vol. 10, pp. 839–850

Abstract

Gradient inversion attacks can reconstruct the victim's private data once they have access to the victim's model and gradient. However, existing research is still immature, and many attacks are conducted in ideal conditions. It is unclear how damaging such attacks really are and how they can be effectively defended. In this paper, we first summarize the current relevant researches and their limitations. Then we design a general gradient inversion attack framework, which can attack both FedSGD and FedAVG. We propose approaches to enhance the label inference and image restoration, respectively. Our approach surpasses the SOTA attacks, by successfully attacking the batches from ImageNet while other methods fail to attack. Finally, we suggest several defense strategies without any utility loss from extensive experiments. We are confirmed that our work makes people aware of the privacy issues and can actively avoid the potential risks.

Authors 7

1. Jiahui Geng

   University of Stavanger

   Affiliation as printed

   Department of Electrical Engineering and Computer Science, Unviersity of Stavanger, Stavanger, Norway

2. Yongli Mou Aachen Chair of Computer Science 5

   RWTH Aachen University

   Affiliation as printed

   Chair of Computer Science 5, Information Systems and Databases, RWTH Aachen University, Aachen, Germany

3. Qing Li

   University of Stavanger

   Affiliation as printed

   Department of Electrical Engineering and Computer Science, Unviersity of Stavanger, Stavanger, Norway

4. Feifei Li Aachen Institutscluster IMA/ZLW und IfU

   RWTH Aachen University

   Affiliation as printed

   Institutscluster IMA/ZLW und IfU, RWTH Aachen University, Aachen, Germany

5. Oya Beyan

   University of Cologne · University Hospital Cologne

   Affiliation as printed

   Faculty of Medicine and University Hospital Cologne, Institute for Medical Informatics, University of Cologne, Köln, Germany

6. Stefan Decker Aachen Chair of Computer Science 5

   RWTH Aachen University

   Affiliation as printed

   Chair of Computer Science 5, Information Systems and Databases, RWTH Aachen University, Aachen, Germany

7. Chunming Rong

   University of Stavanger

   Affiliation as printed

   Department of Electrical Engineering and Computer Science, Unviersity of Stavanger, Stavanger, Norway