BriefGPT.xyz
Aug, 2019
无监督域自适应的伪标注课程
Pseudo-Labeling Curriculum for Unsupervised Domain Adaptation
HTML
PDF
Jaehoon Choi, Minki Jeong, Taekyung Kim, Changick Kim
TL;DR
通过基于密度的聚类算法提出伪标签课程表,通过高密度值子集进行早期训练,在后期使用低密度值的数据子集,进而改进网络生成伪标签的能力,提高模型的训练效果并实现最先进的性能
Abstract
To learn target discriminative representations, using
pseudo-labels
is a simple yet effective approach for
unsupervised domain adaptation
. However, the existence of false
→