BriefGPT.xyz
Apr, 2018
超参数神经网络的损失景观
The loss landscape of overparameterized neural networks
HTML
PDF
Y Cooper
TL;DR
本文研究过参数神经网络的损失函数的数学特性，证明了当神经网络的参数数目大于数据点数目时，全局最小值点的集合通常是一个高维子流形。
Abstract
We explore some
mathematical features
of the
loss landscape
of
overparameterized neural networks
. A priori one might imagine that the loss
→