多篇顶会看个体因果推断（ITE）的前世今生

背景

预测行动间的因果关系是一个非常重要的研究课题。例如，医生判断哪种药物会对病人产生更好的效果。在这种任务中，我们的可观测数据有过去采取的行动（吃了哪些药），它们的结果（病情），可能还有更多的协变量covariate information（病人信息），但我们不知道行为与结果之间的因果关系。一种简单的估计方法是基于干预，即保证两次实验环境完全一致，只改变 $T$ （是否用药），然后将得到的结果求差即用药的收益。

本文从第一篇以深度学习的工具研究individual treatment effect 的文章开始，挑选了6篇顶会文章对该领域的发展做一介绍。具体包括：

ICML 2017 Estimating individual treatment effect: generalization bounds and algorithms
NeurIPS 2019 Adapting Neural Networks for the Estimation of Treatment Effects
PNAS 2019 Meta-learners for Estimating Heterogeneous Treatment Effects using Machine Learning
AAAI 2020 Learning Counterfactual Representations for Estimating Individual Dose-Response Curves
ICLR 2021 Oral: VCNet and Functional Targeted Regularization For Learning Causal Effects of Continuous Treatments
Arxiv 2021 Neural Counterfactual Representation Learning for Combinations of Treatments

内容中包含的图片若涉及版权问题，请及时与我们联系删除

多篇顶会看个体因果推断（ITE）的前世今生

背景

ICML 2017 Estimating individual treatment effect: generalization bounds and algorithms

NeurIPS 2019 Adapting Neural Networks for the Estimation of Treatment Effects

PNAS 2019 Meta-learners for Estimating Heterogeneous Treatment Effects using Machine Learning

AAAI 2020 Learning Counterfactual Representations for Estimating Individual Dose-Response Curves

ICLR 2021 Oral: VCNet and Functional Targeted Regularization For Learning Causal Effects of Continuous Treatments

Arxiv 2021 Neural Counterfactual Representation Learning for Combinations of Treatments

评论