Understanding Private Evolution as Learning-Augmented Clustering Apple researchers Audra McMillan, Kunal Talwar and Felix Zhou published a paper in November 2026 recasting Private Evolution (PE), a differentially private synthetic data generation algorithm, as generative model-augmented Wasserstein learning. The authors show theoretically that when the generator captures something about the true distribution, sample complexity depends on intrinsic rather than ambient dimension, and they propose a new geometry-aware version of PE with provable convergence on well-clustered instances after showing standard PE variants can fail to converge on such cases. Experimentally, the new algorithm is competitive with standard baselines and can improve recall. content type paper https://machinelearning.apple.com/research/ published November 2026 Understanding Private Evolution as Learning-Augmented Clustering AuthorsAudra McMillan, Kunal Talwar, Felix Zhou† Private Evolution PE is a differentially private algorithm for synthetic data generation. While it can be viewed as a Wasserstein learning algorithm, it performs much better in practice than worst-case Wasserstein analyses would predict. We recast PE as generative model-augmented Wasserstein learning. We show theoretically that when we take into account the use of a generative model that is able to capture something about the true distribution, then we can obtain much better performance bounds. For example, if the generator gives samples in the same low-dimensional space as the distribution, then sample complexity depends on intrinsic, not ambient, dimension. We also show that standard variants of PE can fail to converge on simple well-clustered instances, and propose a new geometry-aware version of PE with provable convergence on such instances. Experimentally, we show that our new algorithm is competitive with standard baselines and can improve recall. Instance-Optimal Private Density Estimation in the Wasserstein Distance November 21, 2024 research area Privacy https://machinelearning.apple.com/research/?domain=Privacy conference NeurIPS https://machinelearning.apple.com/research/?event=NeurIPS Estimating the density of a distribution from samples is a fundamental problem in statistics. In many practical settings, the Wasserstein distance is an appropriate error metric for density estimation. For example, when estimating population densities in a geographic region, a small Wasserstein distance means that the estimate is able to capture roughly where the population mass is. In this work we study differentially private density estimation… Instance Optimal Private Density Estimation in the Wasserstein Distance July 23, 2024 research area Privacy https://machinelearning.apple.com/research/?domain=Privacy conference TPDP https://machinelearning.apple.com/research/?event=TPDP