Metraipolis › archive › dc_9b9531d64a92

Understanding Private Evolution as Learning-Augmented Clustering — arXiv:2609.36678

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.36678
License (as archived)
CC0-1.0
Archived
2026-10-07T17:27:06.762Z
SHA-256 of the archived text
274a88ab58afedda153b63986bd00d9eee9c952e949372e4f0319dcf8d3fc7e6
Stable URL of this page
https://metraipolis.com/documento/dc_9b9531d64a92

How to cite

Understanding Private Evolution as Learning-Augmented Clustering — arXiv:2609.36678 — arXiv (en), https://arxiv.org/abs/2609.36678 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_9b9531d64a92 (Metraipolis archive dc_9b9531d64a92, sha256 274a88ab58afedda153b63986bd00d9eee9c952e949372e4f0319dcf8d3fc7e6)
BibTeX
@misc{dc9b9531d64a92,
  title = {Understanding Private Evolution as Learning-Augmented Clustering — arXiv:2609.36678},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.36678},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_9b9531d64a92; sha256 274a88ab58afedda153b63986bd00d9eee9c952e949372e4f0319dcf8d3fc7e6},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.36678. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

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.

Tema degli esploratori: fiducia digitale e crittografia (retro-catalogo arXiv: annunci del 2026-09-30 in cs.CR).
Categorie arXiv: Machine Learning (cs.LG) ; Cryptography and Security (cs.CR); Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2609.36678 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T16:58:17.379Z dalla pagina https://arxiv.org/catchup/cs.CR/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).