Research map / Learning theory and algorithms / Prediction and decision losses
Auditing Black-Box Prediction Models for Data Minimization Compliance
Bashir Rastegarpanah, Krishna P. Gummadi, Mark Crovella
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use TS, TTTS, Greedy or PM exploration instead of uniform exploration | data minimization auditor (Algorithm 1) · Digits/SVM, Census/DT, Nursery/DT at 1% level | number of system queries | Improved |
| use Greedy and PM instead of Thompson-sampling based TS and TTTS | data minimization auditor (Algorithm 1) · Digits/SVM, Census/DT, Nursery/DT at 1% level | number of system queries | Improved |
| use proposed exploration strategies instead of uniform exploration under a query budget | data minimization measurement (Algorithm 2) · three prediction systems, 0.95 confidence | guaranteed data minimization level | Improved |
| use Greedy and PM instead of TS-based algorithms at small query budgets | data minimization measurement (Algorithm 2) · three prediction systems, small budgets | guaranteed data minimization level | Improved |
| query-based audit instead of population audit | data minimization measurement (Algorithm 2) · three prediction systems | queries needed to reach population-audit level: less than 5% of queries used by population audit | Improved |
| use proposed exploration strategies instead of uniform on a system with an unused extra input | data minimization auditor (Algorithm 1) · Census/DT with unused attribute | number of system queries | Improved |
| use Greedy instead of other exploration strategies | data minimization auditor (Algorithm 1) · Census/DT with unused attribute | number of system queries | Improved |
Closest papers by meaning
- Nearly Tight Black-Box Auditing of Differentially Private Machine Learning (NeurIPS 2024 · similarity 0.85)
- Cost Efficient Fairness Audit Under Partial Feedback (NeurIPS 2026 · similarity 0.85)
- Data Mixture Optimization: A Multi-fidelity Multi-scale Bayesian Framework (NeurIPS 2025 · similarity 0.84)
- SMI: Statistical Membership Inference for Reliable Unlearned Model Auditing (NeurIPS 2026 · similarity 0.84)
- Algorithms with Calibrated Machine Learning Predictions (ICML 2025 · similarity 0.84)
- Random features models: a way to study the success of naive imputation (ICML 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
