• Corpus ID: 235262749

A Critical Look at the Consistency of Causal Estimation with Deep Latent Variable Models

@inproceedings{Rissanen2021ACL,
  title={A Critical Look at the Consistency of Causal Estimation with Deep Latent Variable Models},
  author={Severi Rissanen and Pekka Marttinen},
  booktitle={Neural Information Processing Systems},
  year={2021}
}
Using deep latent variable models in causal inference has attracted considerable interest recently, but an essential open question is their ability to yield consistent causal estimates. While they have demonstrated promising results and theory exists on some simple model formulations, we also know that causal effects are not even identifiable in general with latent variables. We investigate this gap between theory and empirical results with analytical considerations and extensive experiments… 
[PDF] Semantic Reader
13 Citations
Highly Influential Citations
2
Background Citations
8
Methods Citations
4
Results Citations
1

13 Citations

Causal Effect Prediction with Flow-based Inference

Empirical results show that the proposed method outperforms baselines on different datasets and leverages the expressive power of flow-based models and tries to recover the complex relationship between observations and unobserved confounders.

Confounder Balancing for Instrumental Variable Regression with Latent Variable

This paper studies the confounding effects from the unmeasured confounders and the imbalance of observed confounders in IV regression and aims at unbiased causal effect estimation. Recently, nonlinear…

Multi-treatment Effect Estimation from Biomedical Data.

This work proposes a neural network that adopts a multi-task learning approach to estimate the effect of multiple treatments and validated M3E2 in three synthetic benchmark datasets that mimic biomedical datasets.

Adapting to Latent Subgroup Shifts via Concepts and Proxies

This work addresses the problem of unsupervised domain adaptation when the source domain differs from the target domain because of a shift in the distribution of a latent subgroup, and shows how the approach degrades as the size of the shift changes, and verify that it outperforms both covariate and label shift adjustment.

Causal Inference with Conditional Instruments using Deep Generative Models

This paper proposes to learn the represen- tations of the information of a CIV and its conditioning set from data with latent confounders for average causal effect estimation and develops a novel data-driven approach for simultaneously learning the representation of aCIV from measured variables and generating the representations of its conditioningSet given measured variables.

Causal Deep Reinforcement Learning using Observational Data

These deconfounding methods can be combined with the existing model-free DRL algorithms such as soft actor-critic and deep Q-learning, provided that a weak condition can be satisfied by the loss functions of these algorithms.

Sequential Causal Effect Variational Autoencoder: Time Series Causal Link Estimation under Hidden Confounding

This work proposes Sequential Causal Effect Variational Autoencoder (SCEVAE), a novel method for time series causality analysis under hidden confounding based on the CEVAE framework and recurrent neural networks and applies it to synthetic datasets with both linear and nonlinear causal links.

Time Series Causal Link Estimation under Hidden Confounding using Knockoff Interventions

This work proposes to estimate confounded causal links of time series using Sequential Causal Effect Variational Autoencoder (SCEVAE) while applying Knockoff interventions, and demonstrates how using suitable proxy variables improves the causal link estimation in the presence of hidden confounders.

Causal Inference from Small High-dimensional Datasets

Causal-Batle is proposed, a methodology to estimate treatment effects in small high-dimensional datasets in the presence of another high- dimensional dataset in the same feature space and adopts an approach that brings transfer learning techniques into causal inference.

Conceptualizing Treatment Leakage in Text-based Causal Inference

The treatment-leakage problem is defined, the identification as well as the estimation challenges it raises are discussed, and the conditions under which leakage can be addressed by removing the treatment-related signal from the text in a pre-processing step the authors define as text distillation.

References

SHOWING 1-10 OF 28 REFERENCES

Identifying Causal Effects With Proxy Variables of an Unmeasured Confounder.

This work shows that, with at least two independent proxy variables satisfying a certain rank condition, the causal effect is nonparametrically identified, even if the measurement error mechanism, i.e., the conditional distribution of the proxies given the confounder, may not be identified.

Causal Effect Inference with Deep Latent-Variable Models

This work builds on recent advances in latent variable modeling to simultaneously estimate the unknown latent space summarizing the confounders and the causal effect and shows its method is significantly more robust than existing methods, and matches the state-of-the-art on previous benchmarks focused on individual treatment effects.

Measurement bias and effect restoration

  • in causal inference. Biometrika,
  • 2014

Identifying Causal Effect Inference Failure with Uncertainty-Aware Models

A practical approach for integrating uncertainty estimation into a class of state-of-the-art neural network methods used for individual-level causal effect estimation, which enables uncertainty-equipped methods to deal gracefully with situations of "no-overlap", common in high-dimensional data, where standard applications of causal effect approaches fail.

Deep Structural Causal Models for Tractable Counterfactual Inference

The experimental results indicate that the proposed framework can successfully train deep SCMs that are capable of all three levels of Pearl's ladder of causation: association, intervention, and counterfactuals, giving rise to a powerful new approach for answering causal questions in imaging applications and beyond.

Inferring Personalized and Race-Specific Causal Effects of Genomic Aberrations on Gleason Scores: A Deep Latent Variable Model

A joint deep latent variable model (DLVM) is proposed to in silico quantify the personalized and race-specific effects that a genomic aberration may exert on the Gleason Score (GS) of each individual PCa patient, and achieves much higher precision in causal effect inference.

MissDeepCausal: Causal Inference from Incomplete Data Using Deep Latent Variable Models

Inferring causal effects of a treatment, intervention or policy from observational data is central to many applications. However, state-of-the-art methods for causal inference seldom consider the…

The Usual Suspects? Reassessing Blame for VAE Posterior Collapse

It is proved that even small nonlinear perturbations of affine VAE decoder models can produce bad local minima, and in deeper models, analogous minima can force the VAE to behave like an aggressive truncation operator, provably discarding information along all latent dimensions in certain circumstances.

Counterfactual Reasoning for Fair Clinical Risk Prediction

An augmented counterfactual fairness criteria is developed to extend the group fairness criteria of equalized odds to an individual level and provides a means to trade off maintenance of fairness with reduction in predictive performance in the context of a learned generative model.

Adapting Neural Networks for the Estimation of Treatment Effects

A new architecture is proposed, the Dragonnet, that exploits the sufficiency of the propensity score for estimation adjustment, and a regularization procedure is proposed that induces a bias towards models that have non-parametrically optimal asymptotic properties `out-of-the-box`.