面向半监督多标签学习的类别分布感知伪标记（英文版）

英文标题：Class-Distribution-Aware Pseudo Labeling for Semi-Supervised Multi-Label Learning中文摘要：本文提出 Class-distribution-Aware Pseudo Labeling (CAP) 方法，通过设计基于类感知阈值的规则学习框架来控制每个类别的伪标签的数量，并约束伪标签的类别分布接近真实分布，旨在解决半监督多标签学习问题。实验结果证明了 CAP 方法可以有效解决半监督多标签学习问题并提高泛化能力。英文摘要：Pseudo labeling is a popular and effective method to leverage the informationof unlabeled data. Conventional instance-aware pseudo labeling methods oftenassign each unlabeled instance with a pseudo label based on its predictedprobabilities. How