Can I read Enhancing Differential Privacy in Federated Learning on EtoBox?

Enhancing Differential Privacy in Federated Learning by Mohit is a document available to read on EtoBox.

What is Enhancing Differential Privacy in Federated Learning about?

This paper investigates the practicality of incorporating Differential Privacy (DP) in Federated Learning (FL) by tuning the number of local and global iterations. It derives the convergence conditions for the DP-based FedAvg algorithm and demonstrates that while the Laplace mechanism may cause divergence, the Gaussian mechanism can achieve convergence with a fixed number of local iterations. Extensive experiments validate the theoretical findings and provide guidelines for optimizing model accuracy in FL w

Author

Mohit

Language

EN