PulseAugur
中
实时 21:09:45
English(EN) Asymptotic Risk of Learning-Rate Schedules for Least-Squares Regression with Discretely Distributed Features

新研究详细介绍了最小二乘回归的学习率调度器

一篇新发表在arXiv上的研究论文分析了具有离散分布特征的最小二乘回归学习率调度器的渐近风险。该研究聚焦于Slock模型中随机梯度下降(SGD)的最后一个迭代点,推导了常数和Warmup-Stable-Decay(WSD)调度器的精确有限时间风险和渐近等价物。研究结果表明,调整基础学习率会导致 $\Theta(T^{-q})$ 的标度律衰减,其中任何固定相对长度的冷却都会使风险提高一个多项式因子,而线性衰减则最小化常数。 AI

影响 为与机器学习模型训练相关的优化算法提供了理论见解。

排序理由 该条目是发表在arXiv上的学术论文,详细介绍了机器学习领域的理论发现。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv stat.ML 阅读 →

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

新研究详细介绍了最小二乘回归的学习率调度器

本文如何被排名

Signal score
3 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
该条目是发表在arXiv上的学术论文,详细介绍了机器学习领域的理论发现。[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
Same-day
Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

报道来源 [1]

  1. arXiv stat.ML TIER_1 English(EN) · No\'e Colomban, Francis Bach, Fabian Schaipp ·

    具有离散分布特征的最小二乘回归学习率调度器的渐近风险

    arXiv:2610.16518v1 Announce Type: cross Abstract: We study the last iterate of stochastic gradient descent (SGD) on streaming least squares with discretely distributed features (Slock's model), which decouples the dynamics across eigendirections. Under power-law source and capaci…