English(EN) Silent Alignment in Two-Layer Linear Networks: A High-Dimensional Perspective

两层线性网络中观察到静默对齐现象

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-22 04:00

研究人员在高维环境下分析了两层线性神经网络中梯度流的动力学。他们的工作结合了闭式解和随机矩阵理论，提供了权重矩阵随时间变化的精确渐近谱分布。一项关键发现是识别出一种“静默对齐”现象，其中网络的奇异方向快速对齐，而其奇异值和经验损失在稍后的对数时间尺度上才发生变化，此时可见的学习才开始。 AI

影响这项研究为理解神经网络中的学习动力学提供了理论框架，可能为未来的模型设计和训练策略提供信息。

排序理由该集群包含一篇学术论文，详细介绍了机器学习中的一项新理论发现。[lever_c_demoted from research: ic=1 ai=1.0]

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

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; 该集群包含一篇学术论文，详细介绍了机器学习中的一项新理论发现。[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: 3 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。

arXiv cs.LG TIER_1 English(EN) · Simon Martin (DI-ENS, LPENS, SIERRA), Francis Bach (DI-ENS, SIERRA), Giulio Biroli (LPENS) · 2026-10-22 04:00

Silent Alignment in Two-Layer Linear Networks: A High-Dimensional Perspective

arXiv:2610.29864v1 Announce Type: cross Abstract: We study gradient flow in a two-layer linear network in the high-dimensional regime where the input, inner and output dimensions diverge proportionally. Under whitened inputs and isotropic imbalance, we combine a closed-form solut…