.png){kind=icon}
Return
Do Gradient Inversion Attacks Make Federated Learning Unsafe?
.png){kind=icon}
OA
AI
DOI：10.1109/TMI.2023.3239391.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Federated learning (FL) allows the collaborative training of AI models without needing to share raw data. This capability makes it especially interesting for healthcare applications where patient and data privacy is of utmost concern. However, recent works on the inversion of deep neural networks from model gradients raised concerns about the security of FL in preventing the leakage of training data. In this work, we show that these attacks presented in the literature are impractical in FL use-cases where the clients' training involves updating the Batch Normalization (BN) statistics and provide a new baseline attack that works for such scenarios. Furthermore, we present new ways to measure and visualize potential data leakage in FL. Our work is a step towards establishing reproducible methods of measuring data leakage in FL and could help determine the optimal tradeoffs between privacy-preserving techniques, such as differential privacy, and model accuracy based on quantifiable metrics.
Keywords:
Deep Learning
gradient inversion
federated learning
patient privacy
security
Journal
IF:
9.8
Papers:
6.3K
Citations:
3.7W
Organization
Cited Papers
Parkinson's disease: Nigral receptor changes support peptidergic role in nigrostriatal modulation
PREAI
ShareSaveShareSaveShareSaveShareSaveFederated learning in medicine: facilitating multi-institutional collaborations without sharing patient data.png)
SCIENTIFIC REPORTS
OAAI
SCIENTIFIC REPORTSIF3.9
OAAIShareSaveShareSave
View more