Open DP: A Proposal for an Open-Source Suite of Differential Privacy Tools
Talkat Data Privacy: From Foundations to Applications (Simons Institute workshop, 2019) · 2019-03-04..03-08·
summary
Proposal to build an open-source, community‑managed suite called OpenDP that supplies a library of vetted differentially private algorithms, APIs for both novice and data‑scientist users, and deployment tools designed to ship code to the data for repositories, government releases, and company research shares. Plan includes formal verification and human vetting, benchmark datasets, scalable backends (interactive and batch), and a funded core team to maintain audits, documentation, and training; major risks are vetting scalability, funding, and avoiding naive, unsafe default uses.
AI-written summary · about
chapters (10)
- 0:00Introduction and vision
- 1:53Use cases for DP tools
- 5:53Scope of the suite
- 6:50Challenges in current systems
- 9:35Architecture and design
- 14:05Team and organizational needs
- 15:36Discussion on implementation
- 20:00Market dynamics and adoption
- 25:50Vetting and educational needs
- 35:35Conclusion and next steps
related talks
19mPractical Considerations for Differential Privacy
13mEvaluating the Usability of Differential Privacy Tools with Data Practitioners
17mPrivacy in Theory, Bugs in Practice: Grey-Box Testing for Differential Privacy Libraries
01h 00mPANEL: Statistical Theory, Privacy and Data Analysis
11mDesigning an Open-Source Platform for Differentially Private Analytics That Is Usable...
26mAaron Roth, Associate Professor, University of Pennsylvania, NYC 2017
10mCompiling Python Programs into Differentially Private Ones
01h 00mPractical Difficulties in the Application of Differential Privacy for the Release of Large-Scale...