利用多标签评估度量的反单调性归纳多标签规则
机器学习
2020-12-09 v1 机器学习
摘要
利用标签间依赖关系被认为对多标签分类至关重要。规则能够以人类可理解且可解释的方式揭示标签间的依赖关系，如蕴含、包含或排斥。然而，头部含多个标签的规则归纳尤其具有挑战性，因为每条规则需考虑的标签组合数量随可用标签数呈指数增长。为克服此限制，穷举规则挖掘算法通常利用反单调性或可分解性等性质来剪枝搜索空间。在本文中，我们考察常用多标签评估度量是否满足这些性质，从而适于对多标签头部进行搜索空间剪枝。
引用
@article{arxiv.1812.06833,
title = {Exploiting Anti-monotonicity of Multi-label Evaluation Measures for Inducing Multi-label Rules},
author = {Michael Rapp and Eneldo Loza Mencía and Johannes Fürnkranz},
journal= {arXiv preprint arXiv:1812.06833},
year = {2020}
}
备注
Preprint version. To appear in: Proceedings of the Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD) 2018. See http://www.ke.tu-darmstadt.de/bibtex/publications/show/3074 for further information. arXiv admin note: text overlap with arXiv:1812.00050