SCM-VAE: Learning Identifiable Causal Representations via Structural Knowledge
Authors:
- 17 December 2022
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1014-1023
Abstract
The goal of causal representation learning is to map low-level observations to high-level causal concepts to learn interpretable and robust representations for various downstream tasks. Latent variable models such as the variational autoencoder (VAE) are frequently leveraged to learn disentangled representations. However, there are often complex non-linear causal relationships underlying the observed data that cannot be captured through disentangled representations or linear dependence assumptions. Further, an independent conditional prior assumption can make learning causal dependencies in the latent space more challenging. We propose a framework, coined SCM-VAE, which uses apriori causal knowledge, a structural causal prior, and a non-linear additive noise structural causal model (SCM) to learn independent causal mechanisms and identifiable causal representations. We conduct theoretical analysis and perform experiments on synthetic and real-world datasets to show the improved quality of learned causal representations and robustness under interventions.Keywords
Funding Information
- National Science Foundation
- National Institute of General Medical Sciences
- University of Arkansas
This publication has 10 references indexed in Scilit:
- Disentangled Representation with Causal Constraints for Counterfactual Fairness2023
- D’ya Like DAGs? A Survey on Structure Learning and Causal DiscoveryACM Computing Surveys, 2022
- On Pearl’s Hierarchy and the Foundations of Causal Inference2022
- Masked Gradient-Based Causal Structure Learning2022
- CausalVAE: Disentangled Representation Learning via Neural Structural Causal Models2021
- Toward Causal Representation LearningProceedings of the IEEE, 2021
- Deep Learning Face Attributes in the Wild2015
- Equitability, mutual information, and the maximal information coefficientProceedings of the National Academy of Sciences, 2014
- Representation Learning: A Review and New PerspectivesIEEE Transactions on Pattern Analysis and Machine Intelligence, 2013
- Causality2009