Active and Passive Causal Inference Learning

TL;DR

This paper lays out an important set of assumptions that are collectively needed for causal identification, such as exchangeability, positivity, consistency and the absence of interference, and builds out a set of important causal inference techniques by categorizing them into two buckets; active and passive approaches.

Abstract

This paper serves as a starting point for machine learning researchers, engineers and students who are interested in but not yet familiar with causal inference. We start by laying out an important set of assumptions that are collectively needed for causal identification, such as exchangeability, positivity, consistency and the absence of interference. From these assumptions, we build out a set of important causal inference techniques, which we do so by categorizing them into two buckets; active and passive approaches. We describe and discuss randomized controlled trials and bandit-based approaches from the active category. We then describe classical approaches, such as matching and inverse probability weighting, in the passive category, followed by more recent deep learning based algorithms. By finishing the paper with some of the missing aspects of causal inference from this paper, such as collider biases, we expect this paper to provide readers with a diverse set of starting points for further reading and research in causal inference and discovery.

Figures (12)

• Figure 1: Examples of passive and active causal inference methods. Dark gray nodes correspond to observed variables while light gray nodes correspond to latent variables. A square node corresponds to a deterministic variable while a circle corresponds to stochastic variables.
• Figure 2: Population
• Figure 3: Causal Graph
• Figure 5: Generalization through extrapolation
• Figure 6: Generalization through interpolation

Theorems & Definitions (2)

• Example 1
• Example 2: Simpson's paradox illustration Carlson2019