Privacy Leakage in Federated Home Applications Using Gradient Inversion Algorithms

Chen, Wenzhi; Sun, Hongjian; You, Minglei; Jiang, Jing

Abstract

With advances in smart metering infrastructure, household electricity metering data are remotely collected, leading to concerns about household privacy leakage. Federated learning is a promising solution because it avoids direct data uploading. However, recent research shows that the gradient of federated learning contains a certain amount of private information that can be recovered using gradient inversion algorithms. This paper proposes an explainable algorithm to locate the key factors affecting privacy leakage and vulnerable data in home application scenarios. Simulations show comparative results of privacy leakages in different situations and reveal that for home Artificial Intelligence applications, smaller batch sizes, training iterations, and extreme values are prone to causing privacy leaks. Based on that, the advice for protecting federated learning privacy under gradient inversion algorithms is summarized.

Citation

Chen, W., Sun, H., You, M., & Jiang, J. (2024, March). Privacy Leakage in Federated Home Applications Using Gradient Inversion Algorithms. Presented at 2024 International Conference on Industrial Technology (ICIT), Bristol, UK

• Presentation Conference Type: Edited Proceedings
• Conference Name: 2024 International Conference on Industrial Technology (ICIT)
• Start Date: Mar 25, 2024
• End Date: Mar 27, 2024
• Acceptance Date: Mar 25, 2024
• Online Publication Date: Jun 5, 2024
• Publication Date: Mar 25, 2024
• Deposit Date: Jun 24, 2024
• Publisher: Institute of Electrical and Electronics Engineers
• Peer Reviewed: Peer Reviewed
• Series ISSN: 2643-2978
• Book Title: 2024 IEEE International Conference on Industrial Technology (ICIT)
• ISBN: 979-8-3503-4027-3
• DOI: https://doi.org/10.1109/ICIT58233.2024.10540758
• Public URL: https://nottingham-repository.worktribe.com/output/36294327
• Publisher URL: https://ieeexplore.ieee.org/document/10540758