What All We've Been Up To at Knexus Research
Talkat Data Privacy: From Foundations to Applications (Simons Institute workshop, 2019) · 2019-03-04..03-08·
summary
Knexus Research presents three concrete efforts: a DARPA/Brandeis crisis detector that builds a social‑network generator and synthetic caller data, develops noise‑resistant decision metrics and achieves down to epsilon 0.5 no false detection by modeling two‑spike calling behavior; a Census/Simpson evaluation suite that runs randomized three‑marginal metric and classifier propensity‑score tests plus full K‑marginal scans to find where synthesis fails; and a NIST/TopCoder competition workflow with $80,000 prize and a provisional vs sequestered datasets format to validate DP synthetic data.
AI-written summary · about
chapters (13)
- 0:00Intro
- 10:20What is Synthetic Data
- 17:40Open Problems
- 20:10Differential Privacy Challenge
- 23:20What is the Challenge
- 24:53Challenges
- 30:16Data
- 30:51Score
- 31:24Schema
- 32:02Real Data
- 32:47Questions
- 33:34NIST
- 35:38Winners Announcement
related talks
13mDPSynth: From Research to Production—Engineering Differentially Private Synthetic...
10mWhen Privacy Guarantees Meet Pre-Trained LLMs: A Case Study in Synthetic Data
28mInsights, Not Secrets: How Synthetic Data Empowers Privacy
20mSynthetic Data for Privacy, AI Fairness and Democratizing Data Access
29mDo I need to be Dr. Frankenstein to create real-ish synthetic data?
16mCompute Engine Testing with Synthetic Data Generation
01h 43mPerspective
01h 11mCPDP ACADEMIC SESSION III