Proceedings

Efficient and Rigorous Model-Agnostic Explanations

Joao Marques-Silva, Jairo A. Lefebre-Lobaina, Maria Vanina Martinez

Download BibTeX Cite

Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence. Main Track. Pages 2637-2646.

https://doi.org/10.24963/ijcai.2025/294

Keywords

  • Constraint Satisfaction and Optimization: CSO: Constraint learning and acquisition
  • Constraint Satisfaction and Optimization: CSO: Satisfiabilty
  • AI Ethics, Trust, Fairness: ETF: Explainability and interpretability
  • Constraint Satisfaction and Optimization: CSO: Constraint programming

Abstract

Explainable artificial intelligence (XAI) is at the core of trustworthy AI. The best-known methods of XAI are sub-symbolic. Unfortunately, these methods do not give guarantees of rigor. Logic-based XAI addresses the lack of rigor of sub-symbolic methods, but in turn it exhibits some drawbacks. These include scalability, explanation size, but also the need to access the details of the machine learning model. Furthermore, access to the details of an ML model may reveal sensitive information. This

paper builds on recent work on symbolic model-agnostic XAI, which is based on explaining samples of behavior of a blackbox ML model, and proposes efficient algorithms for the computation of explanations. The experiments confirm the scalability of the novel algorithms.

Joao Marques-Silva, Jairo A. Lefebre-Lobaina, and María Vanina Martínez. 2025. Efficient and Rigorous Model-Agnostic Explanations. In Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence, IJCAI-25. IJCAI, 2637–2646. https://doi.org/10.24963/ijcai.2025/294
Welcome Future Conferences AI Journal Division IJCAI Proceedings About IJCAI IJCAI Awards