Deep Structural Causal Models for Tractable Counterfactual Inference (2020-06-11T00:00:00.000000Z)

TL;DR

The experimental results indicate that the proposed framework can successfully train deep SCMs that are capable of all three levels of Pearl's ladder of causation: association, intervention, and counterfactuals, giving rise to a powerful new approach for answering causal questions in imaging applications and beyond.

Abstract

We formulate a general framework for building structural causal models (SCMs) with deep learning components. The proposed approach employs normalising flows and variational inference to enable tractable inference of exogenous noise variables - a crucial step for counterfactual inference that is missing from existing deep causal learning methods. Our framework is validated on a synthetic dataset built on MNIST as well as on a real-world medical dataset of brain MRI scans. Our experimental results indicate that we can successfully train deep SCMs that are capable of all three levels of Pearl's ladder of causation: association, intervention, and counterfactuals, giving rise to a powerful new approach for answering causal questions in imaging applications and beyond. The code for all our experiments is available at this https URL.

Authors

Daniel Coelho de Castro

2 papers

Nick Pawlowski

4 papers

B. Glocker

1 papers

TL;DR

Abstract

Authors

References63 items

Causality and Econometrics

Counterfactual Generation and Fairness Evaluation Using Adversarially Learned Inference

CausalVAE: Disentangled Representation Learning via Neural Structural Causal Models

CausalVAE: Structured Causal Disentanglement in Variational Autoencoder

Attribute-based Regularization of VAE Latent Spaces

Attribute-based regularization of latent spaces for variational auto-encoders

Normalizing Flows for Probabilistic Modeling and Inference

Learning to synthesise the ageing brain without longitudinal data

PyTorch: An Imperative Style, High-Performance Deep Learning Library

Causality for Machine Learning

Explanation by Progressive Exaggeration

Learning the Difference that Makes a Difference with Counterfactually-Augmented Data

Learning Likelihoods with Conditional Normalizing Flows

Explaining Visual Models by Causal Attribution

Neural Spline Flows

Structured Output Learning with Conditional Generative Flows

Counterfactual Off-Policy Evaluation with Gumbel-Max Structural Causal Models

The seven tools of causal inference, with reflections on machine learning

A Meta-Transfer Objective for Learning to Disentangle Causal Mechanisms

Preventing Failures Due to Dataset Shift: Learning Predictive Models That Transport

Pyro: Deep Universal Probabilistic Programming

Morpho-MNIST: Quantitative Assessment and Diagnostics for Representation Learning

Graphical Generative Adversarial Networks

Variational Message Passing with Structured Inference Networks

Conditional Density Estimation with Bayesian Normalising Flows

Learning Independent Causal Mechanisms

Elements of Causal Inference: Foundations and Learning Algorithms

CausalGAN: Learning Causal Implicit Generative Models with Adversarial Training

Deep IV: A Flexible Approach for Counterfactual Prediction

Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraints

Causal Effect Inference with Deep Latent-Variable Models

Image processing and Quality Control for the first 10,000 brain imaging datasets from UK Biobank

Counterfactual Fairness

Face aging with conditional generative adversarial networks

Image-to-Image Translation with Conditional Adversarial Networks

beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework

Categorical Reparameterization with Gumbel-Softmax

The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables

Image Style Transfer Using Convolutional Neural Networks

InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets

Adversarially Learned Inference

Adversarial Feature Learning

Density estimation using Real NVP

Composing graphical models with neural networks for structured representations and fast inference

Autoencoding beyond pixels using a learned similarity metric

Learning Structured Output Representation using Deep Conditional Generative Models

Variational Inference with Normalizing Flows

Deep Convolutional Inverse Graphics Network

UK Biobank: An Open Access Resource for Identifying the Causes of a Wide Range of Complex Diseases of Middle and Old Age

Adam: A Method for Stochastic Optimization

Conditional Generative Adversarial Nets

Stochastic Backpropagation and Approximate Inference in Deep Generative Models

Intriguing properties of neural networks

A Family of Nonparametric Density Estimation Algorithms

Ageing and the brain

AUTO-ENCODING VARIATIONAL BAYES

GENERATIVE ADVERSARIAL NETS

Causality pairwise inference datasets. Replication Data for: "Learning Functional Causal Models with Generative Neural Networks"

GENOME-WIDE ASSOCIATION STUDIES

Causality : Models , Reasoning , and Inference

Causal diagrams for epidemiologic research.

Gradient-based learning applied to document recognition

Gumbel machinery

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names