A Framework for Feasible Counterfactual Exploration incorporating Causality, Sparsity and Density

TL;DR

The paper tackles the challenge of generating counterfactual explanations that are feasible under causal relationships while remaining sparse and actionable. It introduces a VAE-based counterfactual generator guided by a black-box classifier, enforcing unary and binary causal constraints and jointly optimizing proximity, validity, feasibility, and sparsity. By learning two-dimensional latent manifolds, the approach identifies regions dense with feasible counterfactuals and discriminates them from infeasible ones. Across three benchmark datasets, the method achieves high validity and feasibility, with favorable sparsity and competitive proximity, outperforming several baselines. This work advances practical recourse by ensuring CFs align with domain knowledge and real-world feasibility, enabling more actionable decisions in sensitive domains.

Abstract

The imminent need to interpret the output of a Machine Learning model with counterfactual (CF) explanations - via small perturbations to the input - has been notable in the research community. Although the variety of CF examples is important, the aspect of them being feasible at the same time, does not necessarily apply in their entirety. This work uses different benchmark datasets to examine through the preservation of the logical causal relations of their attributes, whether CF examples can be generated after a small amount of changes to the original input, be feasible and actually useful to the end-user in a real-world case. To achieve this, we used a black box model as a classifier, to distinguish the desired from the input class and a Variational Autoencoder (VAE) to generate feasible CF examples. As an extension, we also extracted two-dimensional manifolds (one for each dataset) that located the majority of the feasible examples, a representation that adequately distinguished them from infeasible ones. For our experimentation we used three commonly used datasets and we managed to generate feasible and at the same time sparse, CF examples that satisfy all possible predefined causal constraints, by confirming their importance with the attributes in a dataset.

Figures (6)

• Figure 1: Illustrative example of Counterfactual Explanations
• Figure 2: Illustrative example of possible feasible CF examples and sparsity
• Figure 3: Illustrative example of possible feasible CF examples and manifold representation
• Figure 4: Architecture
• Figure 5: Manifold