研究人员开发了一种新的近似方法来预测具有固定深度的贝叶斯深度神经网络（MLP）的泛化性能。该方法利用等效Wishart Ansatz来模拟分层经验核的涨落，从而能够在比例宽度范围内进行大偏差分析。该框架将深度网络中的表示学习过程简化为一组标量序参数，并通过识别局部核重整化机制将其扩展到卷积架构。 AI
影响 这项研究为理解和预测深度神经网络的行为提供了一个新的理论框架，可能有助于其设计和优化。
排序理由 该集群包含一篇详细介绍分析深度神经网络新理论方法的论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2605.29684v1 Announce Type: cross Abstract: The scaling limit where both the size of the training set $P$ and the width $N$ of a deep neural network grow at the same rate, the so-called proportional-width regime, has been intensely studied for shallow, single-hidden-layer n…
The scaling limit where both the size of the training set $P$ and the width $N$ of a deep neural network grow at the same rate, the so-called proportional-width regime, has been intensely studied for shallow, single-hidden-layer networks. However, extending these non-perturbative…