Metraipolis › archive › dc_c472dee2f34c

On the Comparison of Optimizers for Imbalanced Learning — arXiv:2610.06421

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.06421
License (as archived)
CC0-1.0
Archived
2026-10-06T09:31:47.066Z
SHA-256 of the archived text
954edb045e12ad404bc82d60d92c4852391a58110b82d8f577e32b63c550d79e
Stable URL of this page
https://metraipolis.com/documento/dc_c472dee2f34c

How to cite

On the Comparison of Optimizers for Imbalanced Learning — arXiv:2610.06421 — arXiv (en), https://arxiv.org/abs/2610.06421 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c472dee2f34c (Metraipolis archive dc_c472dee2f34c, sha256 954edb045e12ad404bc82d60d92c4852391a58110b82d8f577e32b63c550d79e)
BibTeX
@misc{dcc472dee2f34c,
  title = {On the Comparison of Optimizers for Imbalanced Learning — arXiv:2610.06421},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.06421},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c472dee2f34c; sha256 954edb045e12ad404bc82d60d92c4852391a58110b82d8f577e32b63c550d79e},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.06421. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Data imbalance is pervasive in machine learning, from rare words and anomalies to underrepresented patterns in heterogeneous or cross-tabulated data. We study idealized optimizers geometries in continuous time to model small-step training in deep learning. We assume that the source of imbalance is unobserved: the optimizer has only access to the aggregate training loss ignoring the exact contributions of the majority and minority groups. In this setting, we characterize a region where majority losses are optimized regardless of the admissible minority structure. We derive explicit equations of this zone and bounds on the time needed to leave it. These bounds exhibit a milder dependence on minority amplitude for sign, spectral, and Newton descent than for Euclidean gradient descent. Experiments with AdamW and Muon suggest similar advantages over SGD across language, tabular, and image tasks.

Tema degli esploratori: statistica e scienza dei dati (elenco arXiv stat.ML, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2610.06421 — 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 2026-10-06T08:24:13.934Z dalla pagina https://arxiv.org/list/stat.ML/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).