mlsanityData to 14 Oct 2026

DeepMed: Semiparametric Causal Mediation Analysis with Debiased Deep Learning

arXiv 2210.04389 ↗First posted 10 Oct 2022stat.ML cs.LG stat.MEcited by 10

Not on the map. Why?

Abstract (arXiv)

Causal mediation analysis can unpack the black box of causality and is therefore a powerful tool for disentangling causal pathways in biomedical and social sciences, and also for evaluating machine learning fairness. To reduce bias for estimating Natural Direct and Indirect Effects in mediation analysis, we propose a new method called DeepMed that uses deep neural networks (DNNs) to cross-fit the infinite-dimensional nuisance functions in the efficient influence functions. We obtain novel theoretical results that our DeepMed method (1) can achieve semiparametric efficiency bound without imposing sparsity constraints on the DNN architecture and (2) can adapt to certain low dimensional structures of the nuisance functions, significantly advancing the existing literature on DNN-based semiparametric causal inference. Extensive synthetic experiments are conducted to support our findings and also expose the gap between theory and practice. As a proof of concept, we apply DeepMed to analyze two real datasets on machine learning fairness and reach conclusions consistent with previous findings.

Key facts. Cited by 10 papers in the map (data through 14 Oct 2026). Builds on 31 of its 69 references that are in the map. First posted on arXiv 10 Oct 2022.

Builds on

31 of 69 references.

References · 31

  1. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  2. 2019HuggingFace's Transformers: State-of-the-art Natural Language Processingcited by 2,218
  3. 2018On the Spectral Bias of Neural Networkscited by 1,341
  4. 2019Benign Overfitting in Linear Regressioncited by 549
  5. 2017Nonparametric regression using deep neural networks with ReLU activation functioncited by 433
  6. 2019Frequency Principle: Fourier Analysis Sheds Light on Deep Neural Networkscited by 368
  7. 2020Improving Reproducibility in Machine Learning Research (A Report from the NeurIPS 2019 Reproducibility Program)cited by 322
  8. 2013Direct and Indirect Effectscited by 280
  9. 2020Deep Network Approximation for Smooth Functionscited by 164
  10. 2018Adaptivity of deep ReLU network for learning in Besov and mixed smooth Besov spaces: optimal rate and curse of dimensionalitycited by 162
  11. 2017Fair Inference On Outcomescited by 144
  12. 2022Causal Machine Learning: A Survey and Open Problemscited by 110
19 more
  1. 2017Implicit Regularization in Deep Learningcited by 90
  2. 2021Sharp Bounds on the Approximation Rates, Metric Entropy, and n-widths of Shallow Neural Networkscited by 86
  3. 2021It's COMPASlicated: The Messy Relationship between RAI Datasets and Algorithmic Fairness Benchmarkscited by 61
  4. 2020Superpolynomial Lower Bounds for Learning One-Layer Neural Networks using Gradient Descentcited by 58
  5. 2020The Surprising Simplicity of the Early-Time Learning Dynamics of Neural Networkscited by 58
  6. 2020The gap between theory and practice in function approximation with deep neural networkscited by 54
  7. 2020Adversarial Estimation of Riesz Representerscited by 35
  8. 2020Learning Deep ReLU Networks Is Fixed-Parameter Tractablecited by 26
  9. 2021On the Representation of Solutions to Elliptic PDEs in Barron Spacescited by 25
  10. 2019A Potential Outcomes Calculus for Identifying Conditional Path-Specific Effectscited by 19
  11. 2020Estimation error analysis of deep learning on the regression problem on the variable exponent Besov spacecited by 13
  12. 2020An Interventionist Approach to Mediation Analysiscited by 10
  13. 2021Proximal mediation analysiscited by 10
  14. 2019Selective machine learning of doubly robust functionalscited by 9
  15. 2020Rejoinder: On nearly assumption-free tests of nominal confidence interval coverage for causal parameters estimated by machine learningcited by 7
  16. 2020Causal Inference of General Treatment Effects using Neural Networks with A Diverging Number of Confounderscited by 7
  17. 2017On semiparametric estimation of a path-specific effect in the presence of mediator-outcome confoundingcited by 6
  18. 2020Double-robust and efficient methods for estimating the causal effects of a binary treatmentcited by 1
  19. 2020Semiparametric causal mediation analysis with unmeasured mediator-outcome confoundingcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 3; 2024: 1; 2025: 2; 2026: 4.

10 papers.

  1. 2023Causal Mediation Analysis with Multi-dimensional and Indirectly Observed Mediatorscited by 4
  2. 2026On the Asymptotic Inadmissibility of Double Machine Learning Estimators Under Structure-Agnostic Modelscited by 3
  3. 2023Assumption-lean falsification tests of rate double-robustness of double-machine-learning estimatorscited by 2
  4. 2023New √n-consistent, numerically stable higher-order influence function estimatorscited by 1
  5. 2024DNA-SE: Towards Deep Neural-Nets Assisted Semiparametric Estimationcited by 0
  6. 2025A Critical Review of Predominant Bias in Neural Networkscited by 0
  7. 2025Linear Causal Representation Learning by Topological Ordering, Pruning, and Disentanglementcited by 0
  8. 2026Decomposing Discrimination: Causal Mediation Analysis for AI-Driven Credit Decisionscited by 0
  9. 2026MediEncoder: Nonlinear Representation Learning for High-Dimensional Causal Mediation Analysiscited by 0
  10. 2026AI-driven Multimodal Representation Learning for Latent Mediation Structure Discovery of Socioeconomic Disadvantage, Psychosocial Factors, and Cardiometabolic Multimorbidity: Insights from the All of Us Research Programcited by 0