BriefGPT.xyz
Jun, 2024
神经网络如何学习：隐含于随机梯度下降的正则化效应
How Neural Networks Learn the Support is an Implicit Regularization Effect of SGD
HTML
PDF
Pierfrancesco Beneventano, Andrea Pinto, Tomaso Poggio
TL;DR
深度神经网络能够有效地学习目标函数的支持，并且小批量随机梯度下降具有隐式正则化效果，通过学习特征的结构，提高特征的可解释性。
Abstract
We investigate the ability of
deep neural networks
to identify the
support
of the target function. Our findings reveal that
mini-batch sgd
→