Counterfactual explanations made clearer using expert fuzzy knowledge
Diversified and Perceptible Counterfactual Examples Leveraging Expert Knowledge
Artificial Intelligence
Summary
Counterfactual explanations help us understand why an AI made a certain decision by showing the smallest changes needed to get a different outcome. The paper shows that these changes should also make sense to people, so they use expert knowledge expressed in fuzzy, linguistic terms to represent how humans perceive data. This approach creates new constraints that stop some usual methods from working, so the authors developed a new version of a popular explanation method called DiCEf. Their method produces explanations that are easier for people to understand while keeping them simple, diverse, and cost-effective.
What this means in practice
- •For machine learning engineers: Generate user-friendly AI explanations personalized with expert knowledge for real-world datasets.
- •For financial risk analysts: Create clearer and more meaningful counterfactuals for model decisions to support regulatory explanations and audits.
Tested on one dataset.
Authors
Akram Bensalem, Fahima Djelil, Marie-Jeanne Lesot, Gr{é}gory Smits
Abstract
CounterFactual Examples (CFEs) are a cornerstone of eXplainable Artificial Intelligence (XAI), offering local, post hoc, and model-agnostic explanations by identifying minimal input modifications that alter a model's prediction. Yet, in order to be intelligible, these modifications must also be semantically meaningful to the explainee. This paper proposes to integrate knowledge expressed as a fuzzy linguistic vocabulary to represent the explainee's perception and interpretation of the data. The domain induced by this fuzzy vocabulary imposes structural constraints that make the features dependent, preventing the use of gradient-based optimisation methods for CFE generation, e.g., DiCE. The paper proposes a continuous data embedding in this linguistic domain and exploits it to define a variant of DiCE that allows personalisation for the explainee, named DiCEf. As illustrated by experimental results on a real-world dataset, this extension of the DiCE method enables the generation of CFEs that are linguistically perceptible while preserving cost minimality, sparsity, and diversity.