Skip to content
Hot eventLive

Transferable Graph Metanetworks 提出可跨宽度迁移的图元网络

1 reports1 sources3 hr ago updated

Get the story

AI overview

2026-10-02,arXiv 发布一手研究,提出可迁移图元网络(Transferable Graph Metanetworks),旨在让图神经网络的性能跨不同宽度迁移。该方法基于不变性与连续性原则,在最大更新参数化(μP)训练的输入网络上表现最强,鲁棒性达训练宽度的 42 倍,并用无限宽极限理论解释泛化保证与失败原因。

Generated from reports · updated 3 hr ago

Timeline

Follow the coverage from different angles.

Oct 2, 2026
  1. arXiv · Statistics Machine Learning
    可迁移图元网络(Transferable Graph Metanetworks)

    提出可迁移图元网络,通过不变性与连续性原则使性能跨不同宽度网络可迁移。在最大更新参数化(μP)训练的输入网络上表现最强,鲁棒性达训练宽度的 42 倍,并用无限宽极限理论解释了泛化保证与失败原因。

Heat trend

There is not enough continuous observation data to draw a trend yet.