Metraipolis › archive › dc_df7ea0250965

Estimation of the Label-Noise Transition Matrix with Performance Guarantees via Selective Classification — arXiv:2609.39829

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.39829
License (as archived): CC0-1.0
Archived: 2026-10-07T13:47:46.545Z
SHA-256 of the archived text: b17ddea039805eeb29555034b6edb6cfc2370972f85ec4d0370d3dc8e0acbd3a
Stable URL of this page: https://metraipolis.com/documento/dc_df7ea0250965

How to cite

Estimation of the Label-Noise Transition Matrix with Performance Guarantees via Selective Classification — arXiv:2609.39829 — arXiv (en), https://arxiv.org/abs/2609.39829 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_df7ea0250965 (Metraipolis archive dc_df7ea0250965, sha256 b17ddea039805eeb29555034b6edb6cfc2370972f85ec4d0370d3dc8e0acbd3a)

BibTeX

@misc{dcdf7ea0250965,
  title = {Estimation of the Label-Noise Transition Matrix with Performance Guarantees via Selective Classification — arXiv:2609.39829},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.39829},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_df7ea0250965; sha256 b17ddea039805eeb29555034b6edb6cfc2370972f85ec4d0370d3dc8e0acbd3a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39829. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Modern machine learning depends heavily on massive datasets, but obtaining high-quality annotations at scale is often expensive. As a result, learning from noisily-labeled data has become common, making accurate estimation of the label-noise transition matrix crucial. However, existing transition matrix estimators rely on the fragile estimation of class-posteriors and do not provide finite-sample performance guarantees. In this work, we propose a novel methodology to estimate the transition matrix based on one-sided selective classification. This approach bypasses class-posterior estimation, provides finite-sample performance guarantees, and leverages flexible learning methods for binary classification. Moreover, we introduce effective algorithms to implement the proposed methodology and provide their refined finite-sample performance bounds.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.LG).
Categorie arXiv: Machine Learning (stat.ML) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.39829 — 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 2026-10-07T13:18:10.673Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).