Metraipolis › archive › dc_348b2fd1f5d8

Robustifying Asynchronous SGD via Soft Throttling — arXiv:2609.39357

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.39357
License (as archived)
CC0-1.0
Archived
2026-10-07T13:15:04.704Z
SHA-256 of the archived text
27b40b6d8402c2be1e01c5f77397b4e5fef8af47cb38901585336d8030dac7d8
Stable URL of this page
https://metraipolis.com/documento/dc_348b2fd1f5d8

How to cite

Robustifying Asynchronous SGD via Soft Throttling — arXiv:2609.39357 — arXiv (en), https://arxiv.org/abs/2609.39357 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_348b2fd1f5d8 (Metraipolis archive dc_348b2fd1f5d8, sha256 27b40b6d8402c2be1e01c5f77397b4e5fef8af47cb38901585336d8030dac7d8)
BibTeX
@misc{dc348b2fd1f5d8,
  title = {Robustifying Asynchronous SGD via Soft Throttling — arXiv:2609.39357},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.39357},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_348b2fd1f5d8; sha256 27b40b6d8402c2be1e01c5f77397b4e5fef8af47cb38901585336d8030dac7d8},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39357. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Asynchronous SGD is a popular algorithm for distributed learning where each client's gradient update is applied on arrival. This leads to a speed-up, but also an increased vulnerability to attacks, as fast clients can dominate the total update. We introduce Throttle, a Byzantine-robust generalization of asynchronous SGD where the key idea is to exponentially down-weight updates from faster clients by a factor $q$. Both asynchronous SGD ($q=1$) and synchronous Byzantine-robust SGD ($q\to\infty$) correspond to specific settings of Throttle. We provide a theoretical analysis of the convergence rate and validate the robustness to attacks both theoretically and empirically. Remarkably, our experiments show that this down-weighting mechanism can also improve performance over standard asynchronous SGD even in the non-Byzantine setting.

Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-10-01 in cs.DC).
Categorie arXiv: Machine Learning (cs.LG) ; Distributed, Parallel, and Cluster Computing (cs.DC); Optimization and Control (math.OC).
Pagina del lavoro: https://arxiv.org/abs/2609.39357 — 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 2026-10-07T12:46:39.765Z dalla pagina https://arxiv.org/catchup/cs.DC/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).