Expand globally, shrink locally: Discriminant multi-label learning with missing labels

Expand globally, shrink locally: Discriminant multi-label learning with missing labels
复制标题

全局扩展,局部收缩:缺失标签的判别式多标签学习

DOI:
10.1016/j.patcog.2020.107675
复制
发表时间:
2020-04
影响因子:
8
通讯作者:
Songcan Chen
Songcan Chen
中科院分区:
计算机科学1区
文献类型:
--
作者:
Zhongchen Ma;Songcan Chen

文献摘要

参考文献

被引文献

相似文献

在多标签学习中,缺失标签的问题带来了一个重大挑战。许多方法试图利用标签矩阵的低秩结构来恢复丢失的标签。然而,这些方法只利用了全局的低秩标签结构,在一定程度上忽略了局部的低秩标签结构和标签判别信息,为进一步的性能改进留下了空间。在本文中,我们开发了一个简单而有效的判别式多标签学习(DM2L)方法,用于缺失标签的多标签学习。具体来说,我们对来自相同标签的实例的所有预测施加低秩结构(秩的局部收缩),并对来自不同标签的实例的预测施加最大分离结构(高秩结构)(秩的全局扩展)。以这种方式,这些强加的低秩结构可以帮助对局部和全局低秩标签结构进行建模,而强加的高秩结构可以帮助提供更多的潜在可辨别性。我们随后的理论分析也支持这些直觉。此外,我们通过使用核技巧来增强DM2L,并建立一个凹凸目标来学习这些模型,提供了一个非线性扩展。与其他方法相比,我们的方法涉及的假设最少,只有一个超参数。即便如此,大量的实验表明,我们的方法仍然优于国家的最先进的方法。
In multi-label learning, the issue of missing labels brings a major challenge. Many methods attempt to recovery missing labels by exploiting low-rank structure of label matrix. However, these methods just utilize global low-rank label structure, ignore both local low-rank label structures and label discriminant information to some extent, leaving room for further performance improvement. In this paper, we develop a simple yet effective discriminant multi-label learning (DM2L) method for multi-label learning with missing labels. Specifically, we impose the low-rank structures on all the predictions of instances from the same labels (local shrinking of rank), and a maximally separated structure (high-rank structure) on the predictions of instances from different labels (global expanding of rank). In this way, these imposed low-rank structures can help modeling both local and global low-rank label structures, while the imposed high-rank structure can help providing more underlying discriminability. Our subsequent theoretical analysis also supports these intuitions. In addition, we provide a nonlinear extension via using kernel trick to enhance DM2L and establish a concave-convex objective to learn these models. Compared to the other methods, our method involves the fewest assumptions and only one hyper-parameter. Even so, extensive experiments show that our method still outperforms the state-of-the-art methods.
DOI: 10.1109/tasl.2007.913750
发表时间: 2008-02-01
影响因子: --
作者:
Turnbull, Douglas;Barrington, Luke;Lanckriet, Gert
通讯作者: Lanckriet, Gert
通过学习特定于标签的特征来改进缺失标签的多标签分类
DOI: 10.1016/j.ins.2019.04.021
发表时间: 2019-08-01
影响因子: 8.1
作者:
Huang, Jun;Qin, Feng;Huang, Qingming
通讯作者: Huang, Qingming
DOI: 10.1016/j.patcog.2017.09.036
发表时间: 2018-02
期刊: Pattern Recognit.
影响因子: --
作者:
Peng Fei Zhu;Qian Xu;Q. Hu;Changqing Zhang;Hong Zhao
通讯作者: Peng Fei Zhu;Qian Xu;Q. Hu;Changqing Zhang;Hong Zhao
DOI: 10.5555/2789272.2789279
发表时间: 2013-09
期刊: ArXiv
影响因子: --
作者:
Qiang Qiu;G. Sapiro
通讯作者: Qiang Qiu;G. Sapiro
DOI: 10.1109/icdm.2014.125
发表时间: 2014-12
期刊: 2014 IEEE International Conference on Data Mining
影响因子: --
作者:
Linli Xu;Zhen Wang;Zefan Shen;Yubo Wang;Enhong Chen
通讯作者: Linli Xu;Zhen Wang;Zefan Shen;Yubo Wang;Enhong Chen