PulseAugur
中
实时 11:54:13
English(EN) Weak-$L^\alpha$ Noise, Strong Guarantees: Gradient Clipping under Affine Heavy Tails

新理论加强了重尾噪声下梯度裁剪的保证

研究人员为优化算法中的梯度裁剪开发了新的理论保证,特别是在仿射重尾噪声下的光滑凸优化。该研究关注裁剪的SGD(Clip-SGD)及其加速变体Clip-SSTM,在弱-$L^\alpha$ 尾部条件下推导了高概率收敛保证。此条件比之前的矩界限限制更少,允许分析具有临界指数$u^{-\alpha}$阶尾部的分布。研究结果表明,这些方法可以达到$K^{-(\alpha-1)/\alpha}$ 的加性随机率,以及来自梯度相关噪声的额外快速衰减项。论文还分析了$\alpha=2$ 的具体情况,并提供了达到由裁剪偏差决定的邻域的有限时间保证,得到了具有帕累托噪声的数值实验支持。 AI

影响 为机器学习中的鲁棒优化提供了理论基础,可能提高模型训练的稳定性。

排序理由 学术论文,详细介绍了优化算法的理论进展。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.LG 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

新理论加强了重尾噪声下梯度裁剪的保证

本文如何被排名

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
学术论文,详细介绍了优化算法的理论进展。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, other
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
32 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

报道来源 [1]

  1. arXiv cs.LG TIER_1 English(EN) · Alyona Nekrasova, Makar Sidorov, Savelii Chezhegov, Alexander Gasnikov, Margarita Preobrazhenskaia, Igor Preobrazhensky, Eduard Gorbunov ·

    弱 $L^\alpha$ 噪声,强保证:仿射重尾下的梯度裁剪

    arXiv:2610.28107v1 Announce Type: cross Abstract: Gradient clipping is a key tool for obtaining high-probability convergence guarantees for stochastic first-order methods under heavy-tailed noise. Existing analyses typically assume a bounded central moment of order $\alpha\in(1,2…