BriefGPT.xyz
Oct, 2020
利用独立因果机制原理学习鲁棒模型
Learning Robust Models Using The Principle of Independent Causal Mechanisms
HTML
PDF
Jens Müller, Robert Schmier, Lynton Ardizzone, Carsten Rother, Ullrich Köthe
TL;DR
本文提出一种基于独立因果机制(ICM)原理的新梯度学习框架,理论和实验结果表明,这种学习方法可以在训练时利用不同环境之间的分布偏移,使神经网络更加鲁棒,并成功地找到与环境无关的关系,达到了对不稳定环境的泛化能力。
Abstract
Standard
supervised learning
breaks down under
data distribution shift
. However, the principle of independent
causal mechanisms
(ICM, Pete
→