arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1511.03260stat.MLcs.LG
收藏

极值分类的层次谱方法

A Hierarchical Spectral Method for Extreme Classification

Microsoft(微软)

Paul Mineiro, Nikos Karampatziakis

摘要:

极端分类问题是多类和多标签分类问题,其中输出的数量如此之大,以至于直接的策略在统计上和计算上都不可行。处理计算负担的一种策略是通过输出空间的树分解。虽然这通常会导致训练和推理随输出数量呈次线性扩展,但它也会导致统计性能降低。在这项工作中,我们确定了树分解方法的两个缺点,并描述了两个启发式缓解。我们组成这些特征值技术构建的树。最终的结果是一个计算效率高的算法,提供了良好的统计性能的几个极端的数据集。

英文摘要:

Extreme classification problems are multiclass and multilabel classification problems where the number of outputs is so large that straightforward strategies are neither statistically nor computationally viable. One strategy for dealing with the computational burden is via a tree decomposition of the output space. While this typically leads to training and inference that scales sublinearly with the number of outputs, it also results in reduced statistical performance. In this work, we identify two shortcomings of tree decomposition methods, and describe two heuristic mitigations. We compose these with an eigenvalue technique for constructing the tree. The end result is a computationally efficient algorithm that provides good statistical performance on several extreme data sets.

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。

↑