Skip to content

Latest commit

 

History

History
66 lines (48 loc) · 1.22 KB

README.md

File metadata and controls

66 lines (48 loc) · 1.22 KB
Raw

dcaPF-torch

A PyTorch prototype for Difference-of-Convex Privacy Funnel

Accepted for ISIT 2024 Learn to compress workshop

Paper link

Citation

Cite the work if you find it useful

@article{huang2024efficient,
  title={An efficient difference-of-convex solver for privacy funnel},
  author={Huang, Teng-Hui and Gamal, Hesham El},
  journal={arXiv preprint arXiv:2403.04778},
  year={2024}
}

Overview

The source code implements a differnce-of-convex solver for the privacy funnel The algorithm is implemented with a neural network. It is a supervised learning algorithm.

Requirements

  • python == 3.7.12
  • PyTorch == 1.11.0
  • torchvision == 0.12.0
  • umap-learn == 0.5.3
  • pandas == 1.3.5
  • scikit-learn == 1.0.2
  • scipy == 1.7.3
  • matplotlib == 3.5.3
  • cvxpy == 1.5.2

Create a Conda Environment

conda env create -f environment.yml

Run the MNIST example

python train_rt.py mnist

Run the Fashion-MNIST example

python train_rt.py fashion

Load a trained model

python load_test.py path/to/model.pth

Developer

Teng-Hui Huang

Electrical and Computer Engineering

University of Sydney, NSW, Australia

email: tenghui[DOT]huang@sydney[DOT]edu[DOT]au