Deep IV: A Flexible Approach for Counterfactual Prediction

By Jason S. Hartford, Greg Lewis, Kevin Leyton-Brown, Matt Taddy. 2017

362 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Causal Inference & Discovery

Citations in the corpus

Deep Neural Networks for Estimation and Inference: Application to Causal Effects and Other Semiparametric Estimands (2018, background): “Given the renewed interest in deep networks, this is an area of study already (Hartford et al., 2017; Polson and Rockova, 2018) and we expect this to continue and that implementations will rapidly evolve.”
Generalized random forests (2016, background): “In other threads, Darolles et al. (2011) and Newey and Powell (2003) study instrumental variables models that generalize the conditionally linear treatment model and allowing for non-linear effects, and Hartford et al. (2017) develop deep learning tools.”
Forecasting Treatment Responses Over Time Using Recurrent Marginal Structural Networks (2018, methodology): “Deep learning has also been used to estimate individualized treatment effects for a single intervention at a fixed time, using instrumental variable approaches [13], generative adversarial networks [44] and multi-task architectures [3].”
Tackling Climate Change with Machine Learning (2019, background): “Such problems could represent a useful application of deep learning methods for counterfactual prediction [283].”
A Survey on Causal Inference (2020, background): “ntal variables would lead to the different assignment of treatment, which is independent of the latent variables, and this assignment is as good as randomization for the purposes of causal inference. [50] broke A Survey on Causal Inference • 25 instrumental variables analysis into two supervised sta”
Interpretable Almost Matching Exactly With Instrumental Variables (2019, methodology): “…include Bayesian methods (Imbens and Rubin, 1997), semiparametric methods (Abadie, 2003; Tan, 2006; Ogburn et al., 2015), nonparametric methods (Frölich, 2007) and deep learning methods (Hartford et al., 2017), but these methods do not enjoy the benefits of interpretability that matching provides.”
Deep Structural Causal Models for Tractable Counterfactual Inference (2020, background): “Causal DL models could be capable of learning relationships from complex high-dimensional data and of providing answers to interventional and counterfactual questions, although current work on deep counterfactuals is limited by modelling only direct cause-effect relationships [11] or instrumental-va”
Performative Prediction (2020, background): “In particular, consider recent work on counterfactual prediction [17].”