PulseAugur
中
实时 09:38:39
English(EN) Scaling and Renormalization in Neural Networks: An Effective Renormalization-Group Description of Grokking

新框架使用重整化群理论描述神经网络的格罗金现象

研究人员开发了一个新的框架,使用缩放和重整化群(RG)理论来描述深度神经网络中格罗金现象。格罗金的特点是从记忆到泛化的延迟转变。该研究聚焦于具有两层神经网络的模加法,建立了与训练周期和权重衰减作为控制坐标一致的序参量和涨落代理。通过构建精确的耦合矩阵并使用其Fiedler向量定义有效几何,研究人员推导出的缩放指数和临界指数与经验估计值非常吻合。 AI

影响 为理解神经网络中的泛化提供了一个理论框架,可能导致更健壮和高效的模型训练。

排序理由 该集群包含一篇详细介绍理解神经网络动力学新理论框架的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

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

新框架使用重整化群理论描述神经网络的格罗金现象

本文如何被排名

Signal score
14 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
该集群包含一篇详细介绍理解神经网络动力学新理论框架的学术论文。[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
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

  1. arXiv cs.AI TIER_1 English(EN) · Min Chen, Junqi Wang, Zeguan Wu, Wenhao Huang, Junyu Liu ·

    神经网络中的缩放与重整化:Grokking的有效重整化群描述

    arXiv:2610.36113v1 Announce Type: cross Abstract: Grokking is a delayed transition from memorization to generalization in deep neural networks (NNs). We examine whether scaling and renormalization group (RG) theories can describe this dynamics, establishing a quantitative physica…