A Model-Agnostic SAT-based Approach for Symbolic Explanation Enumeration
This work addresses the need for interpretable AI by providing a generic method to explain predictions, though it appears incremental as it builds on existing explanation types.
The paper tackled the problem of generating symbolic explanations for machine learning predictions by proposing a model-agnostic SAT-based approach to produce sufficient reasons and counterfactuals, with experimental results on image classification demonstrating feasibility and effectiveness.
In this paper titled A Model-Agnostic SAT-based approach for Symbolic Explanation Enumeration we propose a generic agnostic approach allowing to generate different and complementary types of symbolic explanations. More precisely, we generate explanations to locally explain a single prediction by analyzing the relationship between the features and the output. Our approach uses a propositional encoding of the predictive model and a SAT-based setting to generate two types of symbolic explanations which are Sufficient Reasons and Counterfactuals. The experimental results on image classification task show the feasibility of the proposed approach and its effectiveness in providing Sufficient Reasons and Counterfactuals explanations.