关于自适应次模函数最小化证明的评注
机器学习
2017-05-11 v1 机器学习
摘要
我们指出“Group-based active query selection for rapid diagnosis in time-critical situations”中出现的定理5的一个问题。定理5限定了贪心算法在常数因子最优范围内识别物品类别的期望查询次数。该定理基于一个关于自适应次模函数最小化的结果的正确性。我们给出一个例子，表明“Adaptive Submodularity: Theory and Applications in Active Learning and Stochastic Optimization”中定理A.11的关键步骤是不正确的。
引用
@article{arxiv.1705.03771,
title = {Comments on the proof of adaptive submodular function minimization},
author = {Feng Nan and Venkatesh Saligrama},
journal= {arXiv preprint arXiv:1705.03771},
year = {2017}
}