On Safeguarding Privacy and Security in the Framework of Federated Learning

By Chuan Ma, Jun Li, Ming Ding, Howard H. Yang, Feng Shu, Tony Q. S. Quek, H. Vincent Poor. 2019

278 citations recorded by the source corpus.

Motivated by the advancing computational capacity of wireless end-user equipment (UE), as well as the increasing concerns about sharing private data, a new machine learning (ML) paradigm has emerged, namely federated learning (FL). Specifically, FL allows a decoupling of data provision at UEs and ML model aggregation at a central unit. By training model locally, FL is capable of avoiding direct data leakage from the UEs, thereby preserving privacy and security to some extent. However, even if raw data are not disclosed from UEs, an individual's private information can still be extracted by some recently discovered attacks against the FL architecture. In this work, we analyze the privacy and security issues in FL, and discuss several challenges to preserving privacy and security when designing FL systems. In addition, we provide extensive simulation results to showcase the discussed issues and possible solutions.

Subfields: Artificial Intelligence, Electrical and Electronic Engineering, Federated & Distributed Learning, Privacy & Security

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Electrical and Electronic Engineering

In Federated & Distributed Learning

In Privacy & Security

Citations in the corpus

Federated Learning With Differential Privacy: Algorithms and Performance Analysis (2019, neutral): “, weights trained in neural networks [21]–[23].”
A review of applications in federated learning (2020, background): “Faced with these vulnerabilities, the existing privacy-preserving methods to enhance privacy guarantees mainly focus on information encryption for client or secure aggregation at server side as well as security protection for FL framework (Ma et al., 2019).”
Federated learning for predicting clinical outcomes in patients with COVID-19 (2021, background): “Governance of data for FL is maintained locally, alleviating privacy concerns, with only model weights or gradients communicated between client sites and the federated server 38,39 .”
Distributed Learning in Wireless Networks: Recent Progress and Future Challenges (2021, background): “, focused only on FL [8]–[18], [21]–[23] or only on MARL [25]), do not discuss the major intertwining between wireless systems and distributed learning”
Federated Machine Learning: Survey, Multi-Level Classification, Desirable Criteria and Future Directions in Communication and Networking Systems (2021, background): “Differential privacy is a mathematical concept that is used in the scope of statistical machine learning to fight against differential attacks and ensure that the processes of collecting, aggregating and analyzing data do not expose sensitive information on individual users [170], [171], [172].”
Federated Learning Meets Blockchain in Edge Computing: Opportunities and Challenges (2021, background): “To this end, the key contributions of this article are highlighted as follows: 1) We provide an overview of the fundamentals of FL and blockchain and propose a novel FLchain architecture applicable to edge computing networks.”
Federated Learning for Smart Healthcare: A Survey (2021, background): “1However, it should be noted that FL cannot fully address the privacy problem in smart healthcare [31].”
A Survey on federated learning* (2020, background): “Faced with these vulnerabilities mentioned above, the existing privacy-preserving methods to enhance privacy guarantees mainly focus on information encryption for client or secure aggregation at server side as well as security protection for FL framework [43].”
A Survey on Federated Learning: The Journey From Centralized to Distributed On-Site Learning and Beyond (2021, background): “information about clients [19], [97], and even worse can induce security issues [98].”
Federated Learning for Internet of Things: A Comprehensive Survey (2021, methodology): “Perturbation techniques [188] such as differential privacy or dummy can be used to protect training datasets against data breach, by constructing composition theorems with complex mathematical solutions.”
Blockchain for 5G and Beyond Networks: A State of the Art Survey (2019, background): “devices while ensuring high learning performance and thus promises to play a significant role in supporting privacysensitive 5G mobile applications such as edge computing and catching, networking, and spectrum management [189].”
Decentralized Federated Learning: A Survey and Perspective (2023, background): “…form the chain of suspicion: 1) whether the server aggregation model is reasonable; 2) whether the global model will have high performance across all clients; 3) whether the global model is validated; and 4) how to use the global model securely, and whether the server is secure from attacks [99].”
Federated Learning for 6G: Applications, Challenges, and Opportunities (2021, background): “However, it is also possible for an eavesdropper to conduct approximate reconstruction of the original data, particularly when the local and global model coefficients cannot be protected [87].”