sghalebikesabi/dpfastmh

DP-Fast MH: Private, Fast, and Accurate Metropolis-Hastings for Large-Scale Bayesian Inference

This repository contains code for the paper DP-Fast MH: Private, Fast, and Accurate Metropolis-Hastings for Large-Scale Bayesian Inference, accepted in International Conference on Machine Learning (ICML), 2023.

@article{zhang2023dpfastmh,
  title={DP-Fast MH: Private, Fast, and Accurate Metropolis-Hastings for Large-Scale Bayesian Inference},
  author={Zhang, Wanrong and Zhang, Ruqi},
  journal={International Conference on Machine Learning},
  year={2023}
}

Usage

Gaussian Mixture

Please run

cd mog
julia dpfastmh.jl

Or

julia dpfastmh_full.jl

Logistic Regression on MNIST

Please run

cd logistic
julia dpfastmh.jl

Or

julia dpfastmh_full.jl