BriefGPT.xyz
Jun, 2024
从标签比例中学习的乐观速率
Optimistic Rates for Learning from Label Proportions
HTML
PDF
Gene Li, Lin Chen, Adel Javanmard, Vahab Mirrokni
TL;DR
弱监督学习问题中的主题是来自标签比例的学习,研究了多种实现分类损失的标准,包括经典的实证比例风险最小化、去偏差的比例平方损失和最近提出的EasyLLP学习规则,这些规则在可实现和不可知设置中均取得了“乐观速率”,并且在样本复杂度上接近最优(log因子)。
Abstract
We consider a
weakly supervised learning
problem called
learning from label proportions
(LLP), where examples are grouped into ``bags'' and only the average label within each bag is revealed to the learner. We st
→