• Corpus ID: 229297802

The Variational Method of Moments

@article{Bennett2020TheVM,
  title={The Variational Method of Moments},
  author={Andrew Bennett and Nathan Kallus},
  journal={ArXiv},
  year={2020},
  volume={abs/2012.09422}
}
The conditional moment problem is a powerful formulation for describing structural causal parameters in terms of observables, a prominent example being instrumental variable regression. A standard approach is to reduce the problem to a finite set of marginal moment conditions and apply the optimally weighted generalized method of moments (OWGMM), but this requires we know a finite set of identifying moments, can still be inefficient even if identifying, or can be unwieldy and impractical if we… 
Quasi-Bayesian Dual Instrumental Variable Regression
TLDR
A novel quasi-Bayesian procedure for IV regression, building upon the recently developed kernelized IV models and the dual/minimax formulation ofIV regression, which produces informative uncertainty estimates on complex high-dimensional problems.
Scalable Quasi-Bayesian Inference for Instrumental Variable Regression
TLDR
This work presents a scalable quasi-Bayesian procedure for IV regression, building upon the recently developed kernelized IV models, and leads to a scalable approximate inference algorithm with time cost comparable to the corresponding point estimation methods.
Adversarial Estimators
TLDR
An asymptotic theory of adversarial estimators (‘A-estimators’), where both the estimator and estimand are defined as the critical points of a sample and population average respectively, is developed and the asymPTotic root-n normality for plug-in estimates of smooth functionals of their parameters is derived.
Causal Inference Under Unmeasured Confounding With Negative Controls: A Minimax Learning Approach
TLDR
This paper tackles the primary challenge to causal inference using negative controls: the identification and estimation of these bridge functions, and provides a new identification strategy that avoids both uniqueness and completeness.
Learning Causal Relationships from Conditional Moment Conditions by Importance Weighting
TLDR
A method is proposed that transforms conditional moment conditions to unconditional moment conditions through importance weighting using the conditional density ratio, and proposes a method that successfully approximates conditional moments conditions.
A Minimax Learning Approach to Off-Policy Evaluation in Confounded Partially Observable Markov Decision Processes
We consider off-policy evaluation (OPE) in Partially Observable Markov Decision Processes (POMDPs), where the evaluation policy depends only on observable variables and the behavior policy depends on
Proximal Reinforcement Learning: Efficient Off-Policy Evaluation in Partially Observed Markov Decision Processes
TLDR
This work considers off-policy evaluation in a partially observed MDP (POMDP) by considering estimating the value of a given target policy in a POMDP given trajectories with only partial state observations generated by a different and unknown policy that may depend on the unobserved state.
Off-policy Evaluation in Infinite-Horizon Reinforcement Learning with Latent Confounders
TLDR
It is shown how, given only a latent variable model for states and actions, policy value can be identified from off-policy data, and optimal balancing can be combined with such learned ratios to obtain policy value while avoiding direct modeling of reward functions.
Towards Principled Causal Effect Estimation by Deep Identifiable Models
TLDR
This work proposes Intact-VAE, a new variant of variational autoencoder (VAE), motivated by the prognostic score that is sufficient for identifying TEs, and proves identiflcation of TEs under unconfoundedness, and also discusses (possible) extensions to harder settings.

References

SHOWING 1-10 OF 18 REFERENCES
Minimax Estimation of Conditional Moment Models
TLDR
This work develops an approach for estimating models described via conditional moment restrictions, and introduces a min-max criterion function, under which the estimation problem can be thought of as solving a zero-sum game between a modeler who is optimizing over the hypothesis space of the target model and an adversary who identifies violating moments over a test function space.
GENERALIZATION OF GMM TO A CONTINUUM OF MOMENT CONDITIONS
This paper proposes a version of the generalized method of moments procedure that handles both the case where the number of moment conditions is finite and the case where there is a continuum of
Efficient Estimation of Models with Conditional Moment Restrictions Containing Unknown Functions
We propose an estimation method for models of conditional moment restrictions, which contain finite dimensional unknown parameters (theta) and infinite dimensional unknown functions (h). Our proposal
Deep Generalized Method of Moments for Instrumental Variable Analysis
TLDR
This paper proposes the DeepGMM algorithm, a new variational reformulation of GMM with optimal inverse-covariance weighting that allows us to efficiently control very many moment conditions and develops practical techniques for optimization and model selection that make it particularly successful in practice.
Adversarial Generalized Method of Moments
TLDR
An approach for learning deep neural net representations of models described via conditional moment restrictions, similar in nature to Generative Adversarial Networks, though here the modeler is learning a representation of a function that satisfies a continuum of moment conditions and the adversary is identifying violating moments.
Asymptotic efficiency in estimation with conditional moment restrictions
Efficient Policy Learning from Surrogate-Loss Classification Reductions
TLDR
This work considers the estimation problem given by a weighted surrogate-loss classification reduction of policy learning with any score function, either direct, inverse-propensity weighted, or doubly robust, and shows that, under a correct specification assumption, the weighted classification formulation need not be efficient for policy parameters.
Breaking the Curse of Horizon: Infinite-Horizon Off-Policy Estimation
TLDR
A new off-policy estimation method that applies importance sampling directly on the stationary state-visitation distributions to avoid the exploding variance issue faced by existing estimators is proposed.
Introduction to Empirical Processes and Semiparametric Inference
Overview.- An Overview of Empirical Processes.- Overview of Semiparametric Inference.- Case Studies I.- Empirical Processes.- to Empirical Processes.- Preliminaries for Empirical Processes.-
Off-policy Evaluation in Infinite-Horizon Reinforcement Learning with Latent Confounders
TLDR
It is shown how, given only a latent variable model for states and actions, policy value can be identified from off-policy data, and optimal balancing can be combined with such learned ratios to obtain policy value while avoiding direct modeling of reward functions.
...
...