BriefGPT.xyz
Jan, 2021
关于主动学习中的统计偏差：如何何时修正
On Statistical Bias In Active Learning: How and When To Fix It
HTML
PDF
Sebastian Farquhar, Yarin Gal, Tom Rainforth
TL;DR
本文探讨了主动学习中样本标注代价昂贵导致的偏差问题并提出了一种新的校正方法，解释了为何忽略这种偏差也能带来一些经验上的成功，尤其对参数众多的神经网络等少量数据训练情境下，这种偏差反而有益。
Abstract
active learning
is a powerful tool when labelling data is expensive, but it introduces a
bias
because the training data no longer follows the population distribution. We formalize this
→