Semiparametric doubly robust targeted double machine learning: a review

2022/03/12 by Edward H. Kennedy, Kennedy, Edward H. · 5 voices · 40 citations
Mathematics · #FOS: Computer and information sciences #Methodology (stat.ME) #stat.ME

paper · pdf · doi:10.48550/arxiv.2203.06469

arxiv published 2022/03/12 · arxiv updated 2023/01/26

Abstract

In this review we cover the basics of efficient nonparametric parameter estimation (also called functional estimation), with a focus on parameters that arise in causal inference problems. We review both efficiency bounds (i.e., what is the best possible performance for estimating a given parameter?) and the analysis of particular estimators (i.e., what is this estimator's error, and does it attain the efficiency bound?) under weak assumptions. We emphasize minimax-style efficiency bounds, worked examples, and practical shortcuts for easing derivations. We gloss over most technical details, in the interest of highlighting important concepts and providing intuition for main ideas.

Cited by

• A Characterization of the Orthocomplement of the Tangent Space of Semiparametric Markov Models
• Linear Regression in a Nonlinear World
• Policy-Aligned Estimation of Conditional Average Treatment Effects
• Assumption-Lean Differential Variance Inference for Heterogeneous Treatment Effect Detection
• Optimal Comprehensible Targeting
• Difference-in-differences with stochastic policy shifts of continuous treatments
• Why Is the Double-Robust Estimator for Causal Inference Not Doubly Robust for Variance Estimation?
• A sensitivity analysis for the average derivative effect
• On the use of cross-fitting in causal machine learning with correlated units
• Machine-Learning-Assisted Comparison of Regression Functions
• Ridge Boosting is Both Robust and Efficient
• Handling Missing Responses under Cluster Dependence with Applications to Language Model Evaluation
• Overlap-weighted orthogonal meta-learner for treatment effect estimation over time
• Estimation of causal dose-response functions under data fusion
• Assessing the robustness of heterogeneous treatment effects in survival analysis under informative censoring
• Control variates for variance-reduced ratio of means estimators
• Doubly Robust Estimation with Stabilized Weights for Binary Proximal Outcomes in Micro-Randomized Trials
• Evaluating and Learning Optimal Dynamic Treatment Regimes under Truncation by Death
• An Orthogonal Learner for Individualized Outcomes in Markov Decision Processes
• Efficient Difference-in-Differences Estimation when Outcomes are Missing at Random
• Overlap-Adaptive Regularization for Conditional Average Treatment Effect Estimation
• Methods for analysing wildlife DNA methylation data
• GDR-learners: Orthogonal Learning of Generative Models for Potential Outcomes
• Rescuing double robustness: safe estimation under complete misspecification
• SADA: Safe and Adaptive Aggregation of Multiple Black-Box Predictions in Semi-Supervised Learning
• Non-overlap Average Treatment Effect Bounds
• Rethinking player evaluation in sports: Goals above expectation and beyond
• A Mathematical Framework for Topological Causal Data Analysis
• Nonparametric efficient estimation of the longitudinal front-door functional
• Estimating Heterogeneous Causal Effect on Networks via Orthogonal Learning
• Everything all at once: On choosing an estimand for multi-component environmental exposures
• Efficient Semiparametric Inference for Distributed Data with Blockwise Missingness
• Covariance as a commutator
• "Within-trial" prognostic score adjustment is targeted maximum likelihood estimation
• Riesz representers for the rest of us
• Distributed Kaplan-Meier Analysis via the Influence Function with Application to COVID-19 and COVID-19 Vaccine Adverse Events
• On Causal Inference for the Survivor Function
• Constructing targeted minimum loss/maximum likelihood estimators: a simple illustration to build intuition
• Efficient nonparametric estimators of discrimination measures with censored survival data
• Propensity score weighting across counterfactual worlds: longitudinal effects under positivity violations

Discussions

• Related to that is the review by Edward Kennedy. arxiv.org/abs/2203.06469 I learned a lot from that (especially after taking the short course at ACIC). [bsky, 4 points, 1 comments]
• A few thoughts: that influence functions obey a chain rule has long been part of the bundle of tricks used to derive them manually arxiv.org/abs/2203.06469 But the hard part is that they are not just [bsky, 3 points, 1 comments]
• Isn't the whole point of the causal machine learning literature that these plug-in estimates are biased? For instance Section 4.1 of arxiv.org/abs/2203.06469 [bsky, 2 points, 4 comments]
• Not textbooks but this may be interesting (?): arxiv.org/abs/2203.06469 and www.tandfonline.com/doi/full/10.... [bsky, 1 points, 1 comments]
• S4.2 of the following gives a more formal description arxiv.org/pdf/2203.06469 [bsky, 0 points, 0 comments]

Related

• Sharp Structure-Agnostic Lower Bounds for General Linear Functional Estimation
• Hyperparameter Tuning for Causal Inference with Double Machine Learning: A Simulation Study
• Minimax Kernel Machine Learning for a Class of Doubly Robust Functionals with Application to Proximal Causal Inference
• It's Hard to Be Normal: The Impact of Noise on Structure-agnostic Estimation
• Selective machine learning of doubly robust functionals