Privacy Enhancing Technologies (PETs)

This repository contains demos and reference implementations for a variety of PETs that will be highlighted at Vector Institute's upcoming PETs Bootcamp in November 2021. Each subdirectory considers one representative example of PET, dataset, and use case.

History

• November 3, 2021 - Demos for PATE and Membership Inference Attacks have been added.
• October 13, 2021 - Demos/notebooks have been added to the repo. Plain Python codebases will be made available later this month.

Accessing Data

During the bootcamp, all datasets (except MIMIC-III) are available at /ssd003/projects/pets/datasets/

For external use, we are providing the following link for downloading datasets:

• OneDrive Link

Please write to [email protected] to request access.

Writing Code

Please do not commit any code directly to the main branch. After cloning the repo, please create a branch for your project / prototype. Please note that any code you push to the repo will be visible by all other bootcamp participants. If you prefer not to contribute code to the bootcamp repo, please download the code and work outside of the repo.

Acknowledgements

Thanks to Vector Institute Applied ML Staff and Interns for their contributions to this work.

• Sara El-Shawa
• Sophie Tian
• Ryan Marten
• Sobhan Hemati
• John Jewell
• Shems Saleh
• Ethan Jackson
• Amrit Krishnan
• Gerald Shen