Adversarial Generalized Method of Moments
arXiv:1803.07164
Abstract
We provide an approach for learning deep neural net representations of models described via conditional moment restrictions. Conditional moment restrictions are widely used, as they are the language by which social scientists describe the assumptions they make to enable causal inference. We formulate the problem of estimating the underling model as a zero-sum game between a modeler and an adversary and apply adversarial training. Our approach is similar in nature to Generative Adversarial Networks (GAN), 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. We outline ways of constructing effective adversaries in practice, including kernels centered by k-means clustering, and random forests. We examine the practical performance of our approach in the setting of non-parametric instrumental variable regression.
Cited by in corpus (14)
- Deep Generalized Method of Moments for Instrumental Variable Analysis
- Auto IV: Counterfactual Prediction via Automatic Instrumental Variable Decomposition
- DoWhy: Addressing Challenges in Expressing and Validating Causal Assumptions
- Causal Inference Under Unmeasured Confounding With Negative Controls: A Minimax Learning Approach
- Biased Stochastic First-Order Methods for Conditional Stochastic Optimization and Applications in Meta Learning
- Instrumental Variable Value Iteration for Causal Offline Reinforcement Learning
- Valid Causal Inference with (Some) Invalid Instruments
- Provably Efficient Neural Estimation of Structural Equation Model: An Adversarial Approach
- On Instrumental Variable Regression for Deep Offline Policy Evaluation
- A Class of Algorithms for General Instrumental Variable Models
- Uplift Modeling from Separate Labels
- Adversarial Machine Learning in Text Analysis and Generation
- Quasi-Bayesian Dual Instrumental Variable Regression
- A general method for regularizing tensor decomposition methods via pseudo-data