Metraipolis › archive › dc_45df475a91d2

Optimal Randomized Proper Online Learning — arXiv:2609.21445

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.21445
License (as archived)
CC0-1.0
Archived
2026-10-10T01:45:05.831Z
SHA-256 of the archived text
7d1d9728fdea2fa282713069eba9ecaf5c6c9c4978c65a2326547e2f286aa9d1
Stable URL of this page
https://metraipolis.com/documento/dc_45df475a91d2

How to cite

Optimal Randomized Proper Online Learning — arXiv:2609.21445 — arXiv (en), https://arxiv.org/abs/2609.21445 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_45df475a91d2 (Metraipolis archive dc_45df475a91d2, sha256 7d1d9728fdea2fa282713069eba9ecaf5c6c9c4978c65a2326547e2f286aa9d1)
BibTeX
@misc{dc45df475a91d2,
  title = {Optimal Randomized Proper Online Learning — arXiv:2609.21445},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.21445},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_45df475a91d2; sha256 7d1d9728fdea2fa282713069eba9ecaf5c6c9c4978c65a2326547e2f286aa9d1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.21445. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

We prove that the optimal expected mistake bound of online learning a function class $\mathcal{H}$ by a randomized proper learning algorithm is $O(\mathtt{L}(\mathcal{H}) \log T)$, where $\mathtt{L}(\mathcal{H})$ is the Littlestone dimension of $\mathcal{H}$ and $T$ is the time horizon. Our result improves upon the previously best known bound of $O(\mathtt{L}(\mathcal{H}) \log^6 T)$ given by Daskalakis and Golowich (STOC 2022), and is optimal up to a universal constant for worst-case classes.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-21 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.21445 — 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 2026-10-10T01:14:16.631Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-21?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).