An open API service indexing awesome lists of open source software.

https://github.com/openmined/pydpvalidator

Validation assets for core OpenMined libraries
https://github.com/openmined/pydpvalidator

Last synced: 3 months ago
JSON representation

Validation assets for core OpenMined libraries

Awesome Lists containing this project

README

          

# PyDPValidator

This repo adapts code from https://github.com/xiyangl3/adp-estimator/ to apply the techniques in that repo's [accompanying paper, Minimax Optimal Estimation of Approximate Differential Privacy on Neighboring Databases](https://arxiv.org/abs/1905.10335) to examining [PyDP](https://github.com/OpenMined/PyDP).
## Dependencies

* To run the experiments, you need to have the following libraries installed:

1. python = 3.6
2. numpy
3. scipy

* The coefficient of best polynomial approximation are pre-computed and stored as ".mat" file. The coefficient of Chebyshev polynomials of the first kind are stored as ".npy" file.

* To get the coefficient of best polynomial approximation, you need to install Chebfun in Matlab through http://www.chebfun.org/

* To get the coefficient of Chebyshev polynomials, you need to install:

1. sympy

## Citing this work

You are encouraged to cite orginal paper for acedamic research:

```bibtex
@inproceedings{liu2019minimax,
title={Minimax Optimal Estimation of Approximate Differential Privacy on Neighboring Databases},
author={Liu, Xiyang and Oh, Sewoong},
booktitle={Advances in Neural Information Processing Systems},
pages={2414--2425},
year={2019}
}
```

## License
[MIT](https://github.com/xiyangl3/adp-estimator/blob/master/LICENSE).