Intact-VAE: 在未观测到的混乱情况下估计治疗效果 (Intact-VAE: Estimating Treatment Effects under Unobserved Confounding)

from arxiv, A major update of the ICLR submission. About 80% of the paper is rewritten, and the theoretical part is totally new. For detailed notes on the update, see https://openreview.net/forum?id=D3TNqCspFpM

As an important problem of causal inference, we discuss the identification and estimation of treatment effects under unobserved confounding. Representing the confounder as a latent variable, we propose Intact-VAE, a new variant of variational autoencoder (VAE), motivated by the prognostic score that is sufficient for identifying treatment effects. We theoretically show that, under certain settings, treatment effects are identified by our model, and further, based on the identifiability of our model (i.e., determinacy of representation), our VAE is a consistent estimator with representation balanced for treatment groups. Experiments on (semi-)synthetic datasets show state-of-the-art performance under diverse settings.

翻译：作为重要的因果推断问题,我们讨论在未观察到的混乱下确定和估计治疗效果的问题。我们代表困惑者作为潜在的变量,提议采用Intact-VAE, 这是一种新的变异自动编码器变异变体,其动机是预测性分数,足以确定治疗效果。我们理论上表明,在某些环境下,治疗效果是由我们的模型确定的,此外,根据我们模型的可识别性(即代表的确定性),我们的VAE是一个一致的估算器,治疗组的代表性是平衡的。关于(半)合成数据集的实验显示了不同环境中的最新表现。