可扩展的多标签分类的标签分布学习
原文中文,约300字,阅读约需1分钟。发表于: 。Scalable Label Distribution Learning (SLDL) is proposed for multi-label classification, where different labels are described as distributions in a latent space with asymmetric correlation,...
本文介绍了一种新颖的任务,即Partial labeling and Long-Tailed Multi-Label Classification(PLT-MLC)。提出了一种端到端的学习框架CO-MIC-Balance,用于解决长尾分布和部分标签的多标签分类问题。实验结果表明,该方法在PLT-MLC数据集上具有更高的有效性和鲁棒性。