2025 ICML ICML 2025

Improving the Variance of Differentially Private Randomized Experiments through Clustering