Enforcing Conditional Independence for Fair Representation Learning and Causal Image Generation

TL;DR

This paper introduces a novel method for enforcing conditional independence in high-dimensional feature spaces, improving fairness and enabling causal image generation with controllable latent representations.

Contribution

It extends CI constraints to high-dimensional features using a dynamic sampling strategy, applicable to any encoder architecture for fair and causal representation learning.

Findings

Achieves high accuracy on downstream tasks while maintaining fairness.

Enables causal image generation with controllable latent spaces.

Effective enforcement of CI in high-dimensional features.

Abstract

Conditional independence (CI) constraints are critical for defining and evaluating fairness in machine learning, as well as for learning unconfounded or causal representations. Traditional methods for ensuring fairness either blindly learn invariant features with respect to a protected variable (e.g., race when classifying sex from face images) or enforce CI relative to the protected attribute only on the model output (e.g., the sex label). Neither of these methods are effective in enforcing CI in high-dimensional feature spaces. In this paper, we focus on a nascent approach characterizing the CI constraint in terms of two Jensen-Shannon divergence terms, and we extend it to high-dimensional feature spaces using a novel dynamic sampling strategy. In doing so, we introduce a new training paradigm that can be applied to any encoder architecture. We are able to enforce conditional independence of the diffusion autoencoder latent representation with respect to any protected attribute under the equalized odds constraint and show that this approach enables causal image generation with controllable latent spaces. Our experimental results demonstrate that our approach can achieve high accuracy on downstream tasks while upholding equality of odds.