arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.08184cs.LGcs.ITmath.ITmath.STstat.MLstat.TH
收藏

对数损失下利用二进制分类器构造多类分类器

Constructing Multiclass Classifiers using Binary Classifiers Under Log-Loss

School of Computer Science and Engineering, Hebrew University of Jerusalem(希伯来大学计算机科学与工程学院)

Assaf Ben-Yishai, Or Ordentlich

摘要:

本文研究了从二进制元素构造多类分类器的方法,并通过后悔度来量化分类器的性能,后悔度是相对于贝叶斯最优对数损失定义的。我们讨论两种已知的方法。第一个是一对所有(OVA),我们证明了多类遗憾的上界的组成分类器的二进制遗憾的总和。第二种是基于二叉树的层次分类。对于这种方法,我们证明了多类遗憾是一个加权和的成分二进制遗憾的权重是由树结构。我们还介绍了一个存储层次分类方法,这可能会产生较小的日志损失和遗憾。这些分类方法的优点是通过模拟合成和现实生活中的数据集。

英文摘要:

The construction of multiclass classifiers from binary elements is studied in this paper, and performance is quantified by the regret, defined with respect to the Bayes optimal log-loss. We discuss two known methods. The first is one vs. all (OVA), for which we prove that the multiclass regret is upper bounded by the sum of binary regrets of the constituent classifiers. The second is hierarchical classification, based on a binary tree. For this method we prove that the multiclass regret is exactly a weighted sum of constituent binary regrets where the weighing is determined by the tree structure. We also introduce a leverage-hierarchical classification method, which potentially yields smaller log-loss and regret. The advantages of these classification methods are demonstrated by simulation on both synthetic and real-life datasets.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑