Relaxed Adaptive Projection

Hello! This GitHub repository contains the source code for the paper Private synthetic data for multitask learning and marginal queries.

Our paper ran experiments on the American Community Survey datasets using the same pre-processing as the Vietri et al. 20 and McKenna et al. 2019 papers.

Requirements and Setup

Our project can be run on CPU and GPU, and the necessary python packages can be installed through pip install -r requirements.txt

Datasets

Datasets can be downloaded from folktables. Our code automatically downloads survey data from 2014. To download the data from a different year, simply passing the argument survey_year=2018 to the data loading function get_acs.

Running the data generator

main.py is the entrypoint for running experiments/generating data.

An example invocation to run an experiment on ACS dataset and specifically using the data for income data in California: python main.py --states CA --tasks income

An example invocation to generate differentially privacy synthetic data on the data for multiple tasks in California: python main.py --states CA --multitask

The default algorithm is RAP++, and we also support RAP in our implementation, see the example below: python main.py --states CA --tasks income --algorithm RAP

To access the script usage, run: python main.py -h

Security

See CONTRIBUTING for more information.

License

This library is licensed under the MIT-0 License. See the LICENSE file.

Citation

Please use the following citation when publishing material that uses our code:

@inproceedings{
vietri2022private,
title={Private Synthetic Data for Multitask Learning and Marginal Queries},
author={Giuseppe Vietri and Cedric Archambeau and Sergul Aydore and William Brown and Michael Kearns and Aaron Roth and Ankit Siva and Shuai Tang and Steven Wu},
booktitle={Advances in Neural Information Processing Systems},
year={2022},
url={https://openreview.net/forum?id=5JdyRvTrK0q}
}

@InProceedings{pmlr-v139-aydore21a,
  title = 	 {Differentially Private Query Release Through Adaptive Projection},
  author =       {Aydore, Sergul and Brown, William and Kearns, Michael and Kenthapadi, Krishnaram and Melis, Luca and Roth, Aaron and Siva, Ankit A},
  booktitle = 	 {Proceedings of the 38th International Conference on Machine Learning},
  year = 	 {2021},
  series = 	 {Proceedings of Machine Learning Research},
  pdf = 	 {http://proceedings.mlr.press/v139/aydore21a/aydore21a.pdf},
  url = 	 {https://proceedings.mlr.press/v139/aydore21a.html},
}