BriefGPT.xyz
May, 2023
学习可分数据的两层神经网络的快速收敛
Fast Convergence in Learning Two-Layer Neural Networks with Separable Data
HTML
PDF
Hossein Taheri, Christos Thrampoulidis
TL;DR
本文探讨了如何在两层神经网络上使用标准化的梯度下降算法，证明了使用标准化梯度下降算法可以实现指数尾数损失函数的快速收敛，并讨论了凸性目标的归纳性及过拟合问题。
Abstract
normalized gradient descent
has shown substantial success in speeding up the convergence of
exponentially-tailed loss functions
(which includes exponential and logistic losses) on linear classifiers with separabl
→