English(EN) Phase Transition from Random Features to Fully Trainable Networks

在神经网络训练动力学中观察到相变

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-23 04:00

研究人员在部分可训练神经网络的近似界限和训练误差中识别出一种相变模式。这些网络结合了固定的随机特征和可适应的表示，显示出准确率的快速初始提升，随后收益递减。该研究为两层网络建立了理论近似率，并通过数值实验验证了这些假设，证明了少量可训练神经元可以捕获大部分准确率提升。 AI

影响为神经网络的训练动力学提供了理论见解，可能为未来的架构设计提供信息。

排序理由关于神经网络属性的理论和数值分析的学术论文。[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: 4 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 关于神经网络属性的理论和数值分析的学术论文。[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。

arXiv cs.LG TIER_1 English(EN) · Xin He, Huan Lao, Xiliang Lu, Shijun Zhang · 2026-10-23 04:00

Phase Transition from Random Features to Fully Trainable Networks

arXiv:2610.31663v1 Announce Type: cross Abstract: Constructing accurate and computationally tractable approximation spaces is a fundamental task in scientific computing. Partially trainable neural networks bridge fixed random features and fully adaptive representations. This pape…