U-CECE: A Universal Multi-Resolution Framework for Conceptual Counterfactual Explanations

TL;DR

U-CECE introduces a flexible, multi-resolution framework for conceptual counterfactual explanations that balances expressivity and efficiency across atomic, relational, and structural levels, adaptable to data and compute constraints.

Contribution

It presents a unified, model-agnostic approach supporting multiple levels of explanation detail, including novel GNN and GAE-based modes for structural explanations.

Findings

Structural counterfactuals are semantically equivalent to ground-truth explanations.

The framework effectively balances efficiency and expressivity across datasets.

Human and LVLM evaluations favor structural explanations over exact GED ground truths.

Abstract

As AI models grow more complex, explainability is essential for building trust, yet concept-based counterfactual methods still face a trade-off between expressivity and efficiency. Representing underlying concepts as atomic sets is fast but misses relational context, whereas full graph representations are more faithful but require solving the NP-hard Graph Edit Distance (GED) problem. We propose U-CECE, a unified, model-agnostic multi-resolution framework for conceptual counterfactual explanations that adapts to data regime and compute budget. U-CECE spans three levels of expressivity: atomic concepts for broad explanations, relational sets-of-sets for simple interactions, and structural graphs for full semantic structure. At the structural level, both a precision-oriented transductive mode based on supervised Graph Neural Networks (GNNs) and a scalable inductive mode based on unsupervised graph autoencoders (GAEs) are supported. Experiments on the structurally divergent CUB and Visual Genome datasets characterize the efficiency-expressivity trade-off across levels, while human surveys and LVLM-based evaluation show that the retrieved structural counterfactuals are semantically equivalent to, and often preferred over, exact GED-based ground-truth explanations.