Learning Joint Nonlinear Effects from Single-variable Interventions in the Presence of Hidden Confounders

We propose an approach to estimate the effect of multiple simultaneous interventions in the presence of hidden confounders. To overcome the problem of hidden confounding, we consider the setting where we have access to not only the observational data but also sets of single-variable interventions in which each of the treatment variables is intervened on separately. We prove identifiability under the assumption that the data is generated from a nonlinear continuous structural causal model with additive Gaussian noise. In addition, we propose a simple parameter estimation method by pooling all the data from different regimes and jointly maximizing the combined likelihood. We also conduct comprehensive experiments to verify the identifiability result as well as to compare the performance of our approach against a baseline on both synthetic and real-world data.

[1]  T. Richardson.  Markov Properties for Acyclic Directed Mixed Graphs , 2003 .

[2]  Frederick Eberhardt,et al.  N-1 Experiments Suffice to Determine the Causal Relations Among N Variables , 2006 .

[3]  D. Floreano,et al.  Revealing strengths and weaknesses of methods for gene network inference , 2010, Proceedings of the National Academy of Sciences.

[4]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[5]  Frederick Eberhardt,et al.  Learning linear cyclic causal models with latent variables , 2012, J. Mach. Learn. Res..

[6]  M. Maathuis,et al.  Estimating high-dimensional intervention effects from observational data , 2008, 0810.4214.

[7]  Xinkun Nie,et al.  Learning Objectives for Treatment Effect Estimation , 2017 .

[8]  Bernhard Schölkopf,et al.  Identifying confounders using additive noise models , 2009, UAI.

[9]  Bernhard Schölkopf,et al.  Causal discovery with continuous additive noise models , 2013, J. Mach. Learn. Res..

[10]  Jin Tian,et al.  A general identification condition for causal effects , 2002, AAAI/IAAI.

[11]  Elias Bareinboim,et al.  Causal Inference by Surrogate Experiments: z-Identifiability , 2012, UAI.

[12]  Yoram Singer,et al.  Adaptive Subgradient Methods for Online Learning and Stochastic Optimization , 2011, J. Mach. Learn. Res..

[13]  Tom Burr,et al.  Causation, Prediction, and Search , 2003, Technometrics.

[14]  Dario Floreano,et al.  Generating Realistic In Silico Gene Networks for Performance Assessment of Reverse Engineering Methods , 2009, J. Comput. Biol..

[15]  N. D. Clarke,et al.  Towards a Rigorous Assessment of Systems Biology Models: The DREAM3 Challenges , 2010, PloS one.

[16]  Bernhard Schölkopf,et al.  On Causal Discovery with Cyclic Additive Noise Models , 2011, NIPS.

[17]  Stefan Wager,et al.  Estimation and Inference of Heterogeneous Treatment Effects using Random Forests , 2015, Journal of the American Statistical Association.

[18]  Esther Duflo,et al.  Generic Machine Learning Inference on Heterogenous Treatment Effects in Randomized Experiments , 2017 .

[19]  J. Pearl.  Causality: Models, Reasoning and Inference , 2000 .

[20]  Aapo Hyvärinen,et al.  On the Identifiability of the Post-Nonlinear Causal Model , 2009, UAI.

[21]  J. Peters,et al.  Identifiability of Gaussian structural equation models with equal error variances , 2012, 1205.2536.

[22]  Elias Bareinboim,et al.  General Identifiability with Arbitrary Surrogate Experiments , 2019, UAI.

[23]  J. Pearl.  Causal diagrams for empirical research , 1995 .

[24]  Xuemin Lin,et al.  A Fast and Effective Heuristic for the Feedback Arc Set Problem , 1993, Inf. Process. Lett..

[25]  Bernhard Schölkopf,et al.  Nonlinear causal discovery with additive noise models , 2008, NIPS.

[26]  M. Maathuis,et al.  Estimating the effect of joint interventions from observational data in sparse high-dimensional settings , 2014, 1407.2451.

[27]  Xinkun Nie,et al.  Quasi-oracle estimation of heterogeneous treatment effects , 2017, Biometrika.

[28]  Bernhard Schölkopf,et al.  Elements of Causal Inference: Foundations and Learning Algorithms , 2017 .

[29]  Daniel Malinsky,et al.  Estimating Causal Effects with Ancestral Graph Markov Models , 2016, Probabilistic Graphical Models.