Paper
@article{angelopoulos-private-sets,
title={Private Prediction Sets},
author={Angelopoulos, Anastasios N and Bates, Stephen and Zrnic, Tijana and Jordan, Michael I},
journal={arXiv preprint arXiv:2102.06202},
year={2021}
}
Basic Overview
This GitHub contains the code we used for the experiments in the private prediction sets paper.
Each experiment lives in a different, appropriately named folder.
The directory core contains code common to all of our experiments, including the implementations of the private quantile subroutine.
The repository is still a work in progress; we will be continually updating the code to make it more user-friendly and remove clutter from our development.
Getting Started
You will need conda.
Execute the following line in the root directory of our repository:
bash setup.sh
This file will set up the conda environment pps and fetch the required data.
Each experiment requires different datasets.
The CIFAR-10 dataset and Coronahack dataset will automatically download when the experiments run or during setup.
For the ./imagenet experiments, you will need to point the scripts towards the val directory of your local copy of the Imagenet dataset.