Hierarchical Variational Autoencoder for Visual Counterfactuals

TL;DR

This paper introduces VAEX, a hierarchical VAE model that improves the generation of visual counterfactuals for explainability in AI by relaxing posterior effects, enabling effective classifier auditing.

Contribution

The paper proposes a novel hierarchical VAE architecture, VAEX, that enhances counterfactual generation for visual data by relaxing posterior effects, advancing explainability methods.

Findings

VAEX successfully generates visual counterfactuals for classifier auditing.

Relaxing the posterior improves counterfactual quality.

Hierarchical structure enhances model flexibility and interpretability.

Abstract

Conditional Variational Auto Encoders (VAE) are gathering significant attention as an Explainable Artificial Intelligence (XAI) tool. The codes in the latent space provide a theoretically sound way to produce counterfactuals, i.e. alterations resulting from an intervention on a targeted semantic feature. To be applied on real images more complex models are needed, such as Hierarchical CVAE. This comes with a challenge as the naive conditioning is no longer effective. In this paper we show how relaxing the effect of the posterior leads to successful counterfactuals and we introduce VAEX an Hierarchical VAE designed for this approach that can visually audit a classifier in applications.