BriefGPT.xyz
Jun, 2021
跨域梯度差异最小化用于无监督域自适应
Cross-Domain Gradient Discrepancy Minimization for Unsupervised Domain Adaptation
HTML
PDF
Zhekai Du, Jingjing Li, Hongzu Su, Lei Zhu, Ke Lu
TL;DR
本文提出了一种跨域梯度分歧最小化方法,使用聚类自监督学习获得目标伪标签,显式地最小化源样本和目标样本生成的梯度差异,以提高目标样本的准确性,实验证明该方法优于许多先前的最新技术
Abstract
unsupervised domain adaptation
(UDA) aims to generalize the knowledge learned from a well-labeled source domain to an unlabeled target domain. Recently,
adversarial domain adaptation
with two distinct classifiers
→