U-CECE: A Universal Multi-Resolution Framework for Conceptual Counterfactual Explanations
2026-04-09 • Artificial Intelligence
Artificial IntelligenceComputer Vision and Pattern Recognition
AI summaryⓘ
The authors address the challenge of explaining AI decisions using concept-based counterfactuals, which either are fast but simplistic or detailed but computationally hard. They introduce U-CECE, a flexible system that adjusts explanation detail from simple concepts to complex graphs depending on available data and computing power. Their method supports different modes for graph explanations and performs well on image datasets, with humans and language models preferring their explanations to traditional complex methods. This helps balance explanation quality and efficiency in AI interpretability.
counterfactual explanationsconcept-based explanationsGraph Edit Distance (GED)graph neural networks (GNNs)graph autoencoders (GAEs)multi-resolution frameworkexplainable AIinterpretabilityVisual Genome datasetCUB dataset
Authors
Angeliki Dimitriou, Nikolaos Chaidos, Maria Lymperaiou, Giorgos Filandrianos, Giorgos Stamou
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.