BriefGPT.xyz
Feb, 2018
基于成对相似度和无标签数据的分类
Classification from Pairwise Similarity and Unlabeled Data
HTML
PDF
Han Bao, Gang Niu, Masashi Sugiyama
TL;DR
提出了一种新的弱监督学习设置，称为SU分类，只需要相似数据对和未标记数据点，可以从SU数据中获得分类风险的无偏估计量，并证明其经验风险最小化器的估计误差达到最优参数收敛速率。通过实验证明了该方法的有效性。
Abstract
One of the biggest bottlenecks in
supervised learning
is its high labeling cost. To overcome this problem, we propose a new weakly-
supervised learning
setting called SU
→