BriefGPT.xyz
Dec, 2008
利用部分标签学习的偏移树
The Offset Tree for Learning with Partial Labels
HTML
PDF
Alina Beygelzimer, John Langford
TL;DR
本文提出了一种叫Offset Tree的算法，用于在只观察到一个选择的收益而不是全部选择的情况下学习决策。该算法将这种情况转化为二元分类，允许在部分信息情况下重用任何现有的完全监督的二元分类算法。实验证明，Offset Tree通常比几种替代方法表现更好，并且是一种在计算上最优的算法，不管是在训练还是测试时，只需要O（log_2 k）的工作即可训练模型或作出预测。
Abstract
We present an algorithm, called the
offset tree
, for
learning
in situations where a loss associated with different decisions is not known, but was randomly probed. The algorithm is an {optimal} reduction from thi
→