Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Spatio-Temporal Graphical Counterfactuals: An Overview

Mingyu Kang, Duxin Chen, Ziyuan Pu, Jianxi Gao, Wenwu Yu

TL;DR

This work presents a survey that compares and discusses different counterfactual models, theories and approaches, and proposes a unified graphical causal framework to infer spatio-temporal counterfactuals.

Abstract

Counterfactual thinking is a crucial yet challenging topic for artificial intelligence to learn knowledge from data and ultimately improve performance for new scenarios. Many research works, including the Potential Outcome Model (POM) and the Structural Causal Model (SCM), have been proposed to address this. However, their modeling, theoretical foundations, and application approaches often differ. Moreover, there is a lack of graphical approaches for inferring spatio-temporal counterfactuals, that account for spatial and temporal interactions among multiple units. Thus, in this work, we aim to present a survey that compares and discusses different counterfactual models, theories and approaches. Additionally, we propose a unified graphical causal framework to infer spatio-temporal counterfactuals.

Spatio-Temporal Graphical Counterfactuals: An Overview

TL;DR

This work presents a survey that compares and discusses different counterfactual models, theories and approaches, and proposes a unified graphical causal framework to infer spatio-temporal counterfactuals.

Abstract

Counterfactual thinking is a crucial yet challenging topic for artificial intelligence to learn knowledge from data and ultimately improve performance for new scenarios. Many research works, including the Potential Outcome Model (POM) and the Structural Causal Model (SCM), have been proposed to address this. However, their modeling, theoretical foundations, and application approaches often differ. Moreover, there is a lack of graphical approaches for inferring spatio-temporal counterfactuals, that account for spatial and temporal interactions among multiple units. Thus, in this work, we aim to present a survey that compares and discusses different counterfactual models, theories and approaches. Additionally, we propose a unified graphical causal framework to infer spatio-temporal counterfactuals.
Paper Structure (16 sections, 14 equations, 11 figures, 2 tables, 1 algorithm)

This paper contains 16 sections, 14 equations, 11 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Potential Outcome Model
  3. Inferring Counterfactuals through POM
  4. Foundational Assumptions of POM
  5. Structural Causal Model
  6. Pearl's Causal Ladder Causality2009APrimer2016why2018pearl
  7. Foundational Assumptions of Bayesian Network
  8. Differences between POM and SCM
  9. Counterfactual Falsifiability
  10. Ignorability and Back-door Criterion
  11. Direct and Indirect Causation
  12. Spatio-Temporal Graphical Counterfactuals
  13. Spatio-Temporal Bayesian Networks
  14. Nonstationarity of STBNs
  15. STBNs and Complex networks
  16. ...and 1 more sections

Figures (11)

  • Figure 1: Towards spatio-temporal graphical counterfactuals.
  • Figure 2: Example of SCM APrimer2016.
  • Figure 3: Diagram of Pearl's causal ladder APrimer2016.
  • Figure 4: Diagram of data extrapolation.
  • Figure 5: Diagram of a back-door path.
  • ...and 6 more figures

Theorems & Definitions (3)

  • Definition 1: d-separation BayesianNetwork1985pearlbelief1986pearlsimulation1987pearlIntelligentSystem1988pearlIdentifyDuLi1990pearl
  • Definition 2: Back-door Criterion diagram1995pearloverview2009pearlCausality2009APrimer2016
  • Definition 3: Front-door Criterion diagram1995pearloverview2009pearlCausality2009APrimer2016