返回全部动态
ReLU 神经网络逼近中参数增长的研究
原标题：On the growth of the parameters of approximating ReLU neural networks
AI 摘要
该论文研究全连接前馈 ReLU 神经网络逼近光滑函数时参数范数的渐近增长问题，而非传统关注宽度或深度的通用逼近性质。主要结论是：对于一类具有当前最优逼近误差的 ReLU 架构，实现参数的范数最多呈多项式增长，且相对于归一化网络规模的增长率在多数情况下优于已有结果，尤其在高维输入时。该结果可用于神经网络训练的误差分析与一致性证明。
以上摘要由 AI 生成，可能存在误差。事实请以原文为准。
正文节选
On the growth of the parameters of a class of approximating ReLU neural networks Abstract This work focuses on the analysis of a class of fully connected feed forward ReLU neural networks as they approximate a given, smooth function. In contrast to conventionally studied universal approximation properties under increasing architectures, e.g., in terms of width or depth of the networks, we are concerned with the asymptotic growth of the norm of the parameters of approximating networks. Such resul
发布时间：2026-11-04 13:00
抓取时间：2026-11-04 13:05
来源机构：arXiv