What We’re Doing in Australia – An API for Confidentialized Contingency Tables
Talkat Data Privacy: From Foundations to Applications (Simons Institute workshop, 2019) · 2019-03-04..03-08·
summary
API lets authorized users generate custom contingency tables (counts, plus means/sums where available) against ABS unit‑record data with PII removed, currently trialed on the MADIP longitudinal extract with ~15 users. Answers use a deterministic per-record perturbation tied to a record key, with zeros left unchanged and a limited per-user query budget to cap risk. Results export to geo formats usable by nationalmap. Open issues noted: joins, weights, longitudinal fields, useful utility with ~100 categorical fields, and how to set a global risk budget.
AI-written summary · about
chapters (12)
- 0:00Intro
- 1:47Where I work
- 3:24Working with the ABS
- 4:04Workshops
- 7:19Confidential Contingency Tables
- 9:54The API
- 10:55The ABS
- 11:57Whats Next
- 12:14Demo
- 15:12Why am I here
- 18:19Algorithms
- 20:11Challenges
related talks
20mData Sharing in Modern Australia
19mBeing Intentional: AB Testing and Data Minimization
43mAnalyzing Administrative Data with Privacy Protection in Place
11mDesigning an Open-Source Platform for Differentially Private Analytics That Is Usable...
40mProtecting sensitive data in huge datasets #bbuzz
01h 00mPractical Difficulties in the Application of Differential Privacy for the Release of Large-Scale...
30mSecurely Democratising Data
30mSharing without Sharing: the Case for Data Clean Rooms Powered by Confidential Computing