Metraipolis › archive › dc_cfcfa2703e3f

Dataset Pruning from First Principles: A Label-Free Linear Programming Approach — arXiv:2610.10347

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.10347
License (as archived)
CC0-1.0
Archived
2026-10-08T12:37:33.218Z
SHA-256 of the archived text
e5111544b460679d5de4358eb0678850941d8ccbe5e5d263f8c09cb2e46302f5
Stable URL of this page
https://metraipolis.com/documento/dc_cfcfa2703e3f

How to cite

Dataset Pruning from First Principles: A Label-Free Linear Programming Approach — arXiv:2610.10347 — arXiv (en), https://arxiv.org/abs/2610.10347 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cfcfa2703e3f (Metraipolis archive dc_cfcfa2703e3f, sha256 e5111544b460679d5de4358eb0678850941d8ccbe5e5d263f8c09cb2e46302f5)
BibTeX
@misc{dccfcfa2703e3f,
  title = {Dataset Pruning from First Principles: A Label-Free Linear Programming Approach — arXiv:2610.10347},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.10347},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cfcfa2703e3f; sha256 e5111544b460679d5de4358eb0678850941d8ccbe5e5d263f8c09cb2e46302f5},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.10347. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Dataset pruning reduces a large training set to a representative subset while preserving model performance. Existing geometry-based methods typically assume that nearby points in embedding space share similar properties. Rather than imposing this assumption, we derive geometric selection criteria by reformulating unbiased subset selection as a variance minimization problem. Unbiasedness ensures that unweighted subset averages recover full-dataset averages in expectation, including losses and gradients at fixed model parameters. Specifically, we characterize a family of unbiased subset selection algorithms as a high-dimensional polytope. In this context, minimizing the expected sampling variance is a linear objective. Differences in sampling variance, averaged over rigid motions, admit closed-form pairwise expressions. Because the polytope has high dimension, directly applying standard linear programming is impractical. We instead use these expressions to construct an efficient vertex walk that optimizes an approximation of the variance objective while preserving unbiasedness, yielding a method that requires neither labels nor model training during selection. Across CIFAR-10, MNIST, and CelebA benchmarks, our method matches or exceeds uniform sampling in mean test accuracy at every evaluated budget and outperforms competing geometric methods in several settings, particularly at small selection budgets. Beyond dataset pruning, the same framework reduces stochastic-gradient variance by increasing diversity within mini-batches while keeping the batch size unchanged.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (stat.ML) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.10347 — 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 2026-10-08T12:14:15.465Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).