This is the companion code to the paper Compression with Exact Error Distribution for Federated Learning.
This repo contains the code for all experiments except the experiments on Distributed Discrete Gaussian (DDG) mechanism, which we directly obtained using the original source code here
- We thank Wei-Ning Chen, for providing us code of his paper, which was partially used in our experiments.
- The code for the Langevin is adapted from the code supplementary material of (Vono et al)(https://proceedings.mlr.press/v151/vono22a.html)