BriefGPT.xyz
Apr, 2021
对于亚高斯混合分布上，超参数化最大间隔分类的风险界
Risk Bounds for Over-parameterized Maximum Margin Classification on Sub-Gaussian Mixtures
HTML
PDF
Yuan Cao, Quanquan Gu, Mikhail Belkin
TL;DR
本文研究现代机器学习系统中深度神经网络等通常高度过参数化的现象，探究其在数据生成、最大边界分类器和风险界限等方面的应用并给出改进性结果。
Abstract
Modern
machine learning
systems such as
deep neural networks
are often highly
over-parameterized
so that they can fit the noisy training d
→