Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Survey of Deep Causal Models and Their Industrial Applications

Zongyu Li, Xiaobo Guo, Siwei Qiang

TL;DR

The paper surveys deep causal models based on neural networks for estimating causal effects from counterfactual data, emphasizing representation learning, debiasing, and counterfactual inference. It traces a development timeline from 2016 to 2023 and classifies methods into five categories: learning balanced representations, covariate confounding learning, GAN-based counterfactual simulation, time-series causal estimation, and multi-treatment/continuous-dose models. It then reviews industrial applications across marketing, e-commerce, economics/finance, medicine, education, and other domains, and compiles datasets and public code resources to guide practice. Finally, it offers experimental guidelines and outlines future directions, including interpretability, fairness, uncertainty quantification, and integrated micro/macro data for robust decision-making in industry.

Abstract

The notion of causality assumes a paramount position within the realm of human cognition. Over the past few decades, there has been significant advancement in the domain of causal effect estimation across various disciplines, including but not limited to computer science, medicine, economics, and industrial applications. Given the continous advancements in deep learning methodologies, there has been a notable surge in its utilization for the estimation of causal effects using counterfactual data. Typically, deep causal models map the characteristics of covariates to a representation space and then design various objective functions to estimate counterfactual data unbiasedly. Different from the existing surveys on causal models in machine learning, this review mainly focuses on the overview of the deep causal models based on neural networks, and its core contributions are as follows: 1) we cast insight on a comprehensive overview of deep causal models from both timeline of development and method classification perspectives; 2) we outline some typical applications of causal effect estimation to industry; 3) we also endeavor to present a detailed categorization and analysis on relevant datasets, source codes and experiments.

A Survey of Deep Causal Models and Their Industrial Applications

TL;DR

The paper surveys deep causal models based on neural networks for estimating causal effects from counterfactual data, emphasizing representation learning, debiasing, and counterfactual inference. It traces a development timeline from 2016 to 2023 and classifies methods into five categories: learning balanced representations, covariate confounding learning, GAN-based counterfactual simulation, time-series causal estimation, and multi-treatment/continuous-dose models. It then reviews industrial applications across marketing, e-commerce, economics/finance, medicine, education, and other domains, and compiles datasets and public code resources to guide practice. Finally, it offers experimental guidelines and outlines future directions, including interpretability, fairness, uncertainty quantification, and integrated micro/macro data for robust decision-making in industry.

Abstract

The notion of causality assumes a paramount position within the realm of human cognition. Over the past few decades, there has been significant advancement in the domain of causal effect estimation across various disciplines, including but not limited to computer science, medicine, economics, and industrial applications. Given the continous advancements in deep learning methodologies, there has been a notable surge in its utilization for the estimation of causal effects using counterfactual data. Typically, deep causal models map the characteristics of covariates to a representation space and then design various objective functions to estimate counterfactual data unbiasedly. Different from the existing surveys on causal models in machine learning, this review mainly focuses on the overview of the deep causal models based on neural networks, and its core contributions are as follows: 1) we cast insight on a comprehensive overview of deep causal models from both timeline of development and method classification perspectives; 2) we outline some typical applications of causal effect estimation to industry; 3) we also endeavor to present a detailed categorization and analysis on relevant datasets, source codes and experiments.
Paper Structure (34 sections, 17 equations, 13 figures, 10 tables)

This paper contains 34 sections, 17 equations, 13 figures, 10 tables.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Definitions
  4. Assumptions
  5. Development of Deep Causal Models
  6. A timeline of development
  7. Model classification
  8. Typical Deep Causal Models
  9. Learning balanced representation
  10. Covariates confounding learning
  11. GANs-based counterfactual simulation
  12. Time series causal estimation
  13. Multi-treatment and continuous models
  14. Current Status of Industrial Applications
  15. Marketing applications
  16. ...and 19 more sections

Figures (13)

  • Figure 1: Rubin causal model in randomized controlled trial or Observational data
  • Figure 2: Causal diagram of proxy variables under unobserved confounding
  • Figure 3: Causal diagram of instrumental variables under unobserved confounding
  • Figure 4: Overview of development timeline of classical deep causal models
  • Figure 5: Taxonomy of deep causal models
  • ...and 8 more figures

Theorems & Definitions (9)

  • Definition 1
  • Definition 2
  • Definition 3
  • Definition 4
  • Definition 5
  • Definition 6
  • Definition 7
  • Definition 8
  • Definition 9