{"type":"ScholarlyArticle","title":"On the Computation of Example-Based Abductive Explanations for Random Forests","authors":[{"name":"Gilles Audemard","slug":"gilles-audemard","given":"Gilles","family":"Audemard","affiliation":"Univ. Artois , CNRS , CRIL , F-62300 Lens , France"},{"name":"Jean-Marie Lagniez","slug":"jean-marie-lagniez","given":"Jean-Marie","family":"Lagniez","affiliation":"Univ. Artois , CNRS , CRIL , F-62300 Lens , France"},{"name":"Pierre Marquis","slug":"pierre-marquis","given":"Pierre","family":"Marquis","affiliation":"Univ. Artois , CNRS , CRIL , F-62300 Lens , France; Institut Universitaire de France {audemard ,"},{"name":"Nicolas Szczepanski","slug":"nicolas-szczepanski","given":"Nicolas","family":"Szczepanski","affiliation":"Univ. Artois , CNRS , CRIL , F-62300 Lens , France"}],"abstract":"We show how to define and compute example-based abductive explanations. Such explanations are guaranteed to be 100% correct, fairly general, and persuasive enough since they cover sufficiently many reference instances furnished by the explainee. We prove that the latter coverage condition yields a complexity shift to the second level of the polynomial hierarchy. We present a CEGAR-based algorithm to derive such explanations, and show how to modify it to derive most anchored example-based abductive explanations, i.e., example-based abductive explanations that cover as many reference instances as possible. We also explain how to reduce example-based abductive explanations to get subset-minimal explanations. Experiments in the case of random forest classifiers show that our CEGAR-based algorithm is quite efficient in practice.","doi":"10.24963/ijcai.2024/407","url":"https://ijcai.org/proceedings/2024/407","pdf":"https://ijcai.org/proceedings/2024/0407.pdf","markdown":"https://ijcai.org/proceedings/2024/407.md","bibtex":"https://ijcai.org/proceedings/2024/bibtex/407","volume":{"slug":"2024","title":"Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2024","url":"https://ijcai.org/proceedings/2024/","year":2024,"location":"Jeju, Korea","dates":"3-9 August 2024","editors":"Kate Larson","isbn":"978-1-956792-04-1","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2024"},"section":"Machine Learning","track":"Main Track","number":407,"pages":{"first":3679,"last":3687},"keywords":["Knowledge Representation and Reasoning: KRR: Computational complexity of reasoning","Constraint Satisfaction and Optimization: CSO: Applications","Machine Learning: ML: Explainable/Interpretable machine learning"],"videos":[],"references":[{"position":1,"raw":"[Amgoud and Ben-Naim, 2022] L. Amgoud and J. Ben- Naim. Axiomatic foundations of explainability. In Proc. of IJCAI'22, pages 636-642, 2022.","title":"Axiomatic foundations of explainability","authors":"L Amgoud, J Ben-Naim","year":"2022","venue":"Proc. of IJCAI'22","doi":null,"url":null,"internal_url":"/proceedings/2022/90"},{"position":2,"raw":"[Audemard and Simon, 2009] G. Audemard and L. Simon. Predicting learnt clauses quality in modern SAT solvers. In Proc. of IJCAI'09, pages 399-404, 2009.","title":"Predicting learnt clauses quality in modern SAT solvers","authors":"G Audemard, L Simon","year":"2009","venue":"Proc. of IJCAI'09","doi":null,"url":null,"internal_url":"/proceedings/2009/74"},{"position":3,"raw":"[Audemard et al., 2020] G. Audemard, F. Koriche, and P. Marquis. On tractable XAI queries based on compiled representations. In Proc. of KR'20, pages 838-849, 2020.","title":"On tractable XAI queries based on compiled representations","authors":"G Audemard, F Koriche, P Marquis","year":"2020","venue":"Proc. of KR'20","doi":null,"url":null,"internal_url":null},{"position":4,"raw":"[Audemard et al., 2022] G. Audemard, S. Bellart, L. Bounia, F. Koriche, J.-M. Lagniez, and P. Marquis. Trading com- plexity for sparsity in random forest explanations. In Proc. of AAAI'22, pages 5461-5469. AAAI Press, 2022.","title":"Trading complexity for sparsity in random forest explanations","authors":"G Audemard, S Bellart, L Bounia, F Koriche, J.-M Lagniez, P Marquis","year":"2022","venue":"Proc. of AAAI'22","doi":null,"url":null,"internal_url":null},{"position":5,"raw":"[Audemard et al., 2023] G. Audemard, J.-M. Lagniez, P. Marquis, and N. Szczepanski. On contrastive expla- nations for tree-based classifiers. In Proc. of ECAI'23, volume 372 of Frontiers in Artificial Intelligence and Applications, pages 117-124. IOS Press, 2023.","title":"On contrastive explanations for tree-based classifiers","authors":"G Audemard, J.-M Lagniez, P Marquis, N Szczepanski","year":"2023","venue":"Frontiers in Artificial Intelligence and Applications","doi":null,"url":null,"internal_url":null},{"position":6,"raw":"[Clarke et al., 2003] E. M. Clarke, O. Grumberg, S. Jha, Y. Lu, and H. Veith. Counterexample-guided abstrac- tion refinement for symbolic model checking. J. ACM, 50(5):752-794, 2003.","title":"Counterexample-guided abstraction refinement for symbolic model checking","authors":"E M Clarke, O Grumberg, S Jha, Y Lu, H Veith","year":"2003","venue":"J. ACM","doi":null,"url":null,"internal_url":null},{"position":7,"raw":"[Cooper and Amgoud, 2023] M. C. Cooper and L. Amgoud. Abductive explanations of classifiers under constraints: Complexity and properties. In Proc. of ECAI'23, pages 469-476, 2023.","title":"Abductive explanations of classifiers under constraints: Complexity and properties","authors":"M C Cooper, L Amgoud","year":"2023","venue":"Proc. of ECAI'23","doi":null,"url":null,"internal_url":null},{"position":8,"raw":"[Darwiche and Hirth, 2020] A. Darwiche and A. Hirth. On the reasons behind decisions. In Proc. of ECAI'20, pages 712-720, 2020.","title":"On the reasons behind decisions","authors":"Hirth Darwiche, A Darwiche, A Hirth","year":"2020","venue":"Proc. of ECAI'20","doi":null,"url":null,"internal_url":null},{"position":9,"raw":"[Gorji and Rubin, 2022] N. Gorji and S. Rubin. Sufficient reasons for classifier decisions in the presence of domain constraints. In Proc. of AAAI'22, pages 5660-5667, 2022.","title":"Sufficient reasons for classifier decisions in the presence of domain constraints","authors":"N Gorji, S Rubin","year":"2022","venue":"Proc. of AAAI'22","doi":null,"url":null,"internal_url":null},{"position":10,"raw":"[Gunning, 2019] D. Gunning. DARPA's explainable artifi- cial intelligence (XAI) program. In Proc. of IUI'19, 2019.","title":"DARPA's explainable artificial intelligence (XAI) program","authors":"D Gunning","year":"2019","venue":"Proc. of IUI'19","doi":null,"url":null,"internal_url":null},{"position":11,"raw":"[Huang et al., 2021] X. Huang, Y. Izza, A. Ignatiev, M. C. Cooper, N. Asher, and J. Marques-Silva. Efficient ex- planations for knowledge compilation languages. CoRR, abs/2107.01654, 2021.","title":"Efficient explanations for knowledge compilation languages","authors":"Huang","year":"2021","venue":null,"doi":null,"url":null,"internal_url":null},{"position":12,"raw":"[Ignatiev et al., 2019] A. Ignatiev, N. Narodytska, and J. Marques-Silva. Abduction-based explanations for ma- chine learning models. In Proc. of AAAI'19, pages 1511- 1519, 2019.","title":"Abduction-based explanations for machine learning models","authors":"A Ignatiev, N Narodytska, J Marques-Silva","year":"2019","venue":"Proc. of AAAI'19","doi":null,"url":null,"internal_url":null},{"position":13,"raw":"[Ignatiev et al., 2020] A. Ignatiev, N. Narodytska, N. Asher, and J. Marques-Silva. On relating 'why?' and 'why not?' explanations. CoRR, abs/2012.11067, 2020.","title":"On relating 'why?' and 'why not?' explanations","authors":"A Ignatiev, N Narodytska, N Asher, J Marques-Silva","year":"2020","venue":null,"doi":null,"url":null,"internal_url":null},{"position":14,"raw":"[Izza and Marques-Silva, 2021] Y. Izza and J. Marques- Silva. On explaining random forests with SAT. In Proc. of IJCAI'21, pages 2584-2591, 2021.","title":"On explaining random forests with SAT","authors":"Y Izza, J Marques-Silva","year":"2021","venue":"Proc. of IJCAI'21","doi":null,"url":null,"internal_url":"/proceedings/2021/356"},{"position":15,"raw":"[Karnaugh, 1953] G. Karnaugh. The map method for syn- thesis of combinational logic circuits. AIEE Transactions on Communications and Electronics, 72:593-599, 1953.","title":"The map method for synthesis of combinational logic circuits","authors":"G Karnaugh","year":"1953","venue":"AIEE Transactions on Communications and Electronics","doi":null,"url":null,"internal_url":null},{"position":16,"raw":"[Kenny et al., 2021] E. M. Kenny, C. Ford, M. S. Quinn, and M. T. Keane. Explaining black-box classifiers using post-hoc explanations-by-example: The effect of expla- nations and error-rates in XAI user studies. Artif. Intell., 294:103459, 2021.","title":"Explaining black-box classifiers using post-hoc explanations-by-example: The effect of explanations and error-rates in XAI user studies","authors":"E M Kenny, C Ford, M S Quinn, M T Keane","year":"2021","venue":"Artif. Intell","doi":null,"url":null,"internal_url":null},{"position":17,"raw":"[Miller, 2019] T. Miller. Explanation in artificial intelli- gence: Insights from the social sciences. Artificial Intelli- gence, 267:1-38, 2019.","title":"Explanation in artificial intelligence: Insights from the social sciences","authors":"T Miller","year":"2019","venue":"Artificial Intelligence","doi":null,"url":null,"internal_url":null},{"position":18,"raw":"[Molnar, 2019] Ch. Molnar. Interpretable Machine Learn- ing -A Guide for Making Black Box Models Explainable. Leanpub, 2019.","title":"Interpretable Machine Learning -A Guide for Making Black Box Models Explainable","authors":"Ch Molnar, Molnar","year":"2019","venue":null,"doi":null,"url":null,"internal_url":null},{"position":19,"raw":"[Narodytska et al., 2018] N. Narodytska, S. Prasad Ka- siviswanathan, L. Ryzhyk, M. Sagiv, and T. Walsh. Verify- ing properties of binarized deep neural networks. In Proc. of AAAI'18, pages 6615-6624, 2018.","title":"Verifying properties of binarized deep neural networks","authors":"N Narodytska, S Prasad Kasiviswanathan, L Ryzhyk, M Sagiv, T Walsh","year":"2018","venue":"Proc. of AAAI'18","doi":null,"url":null,"internal_url":null},{"position":20,"raw":"[Nauta et al., 2023] M. Nauta, J. Trienes, S. Pathak, E. Nguyen, M. Peters, Y. Schmitt, J. Schlötterer, M. van Keulen, and Ch. Seifert. From anecdotal evidence to quan- titative evaluation methods: A systematic review on eval- uating explainable ai. ACM Comput. Surv., 55(13s), 2023.","title":"From anecdotal evidence to quantitative evaluation methods: A systematic review on evaluating explainable ai","authors":"M Nauta, J Trienes, S Pathak, E Nguyen, M Peters, Y Schmitt, J Schlötterer, M Van Keulen, Ch Seifert","year":"2023","venue":"ACM Comput. Surv","doi":null,"url":null,"internal_url":null},{"position":21,"raw":"[Poché et al., 2023] A. Poché, L. Hervier, and M. C. Bakkay. Natural example-based explainability: A survey. In Proc. of XAI'23, volume 1902 of Communications in Computer and Information Science, pages 24-47. Springer, 2023.","title":"Natural example-based explainability: A survey","authors":"A Poché, L Hervier, M C Bakkay","year":"1902","venue":"Proc. of XAI'23","doi":null,"url":null,"internal_url":null},{"position":22,"raw":"[Roussel and Manquinho, 2021] O. Roussel and V. M. Man- quinho. Pseudo-boolean and cardinality constraints. In Handbook of Satisfiability -Second Edition, volume 336 of Frontiers in Artificial Intelligence and Applications, pages 1087-1129. IOS Press, 2021.","title":"Pseudo-boolean and cardinality constraints","authors":"O Roussel, V M Manquinho","year":"2021","venue":"Frontiers in Artificial Intelligence and Applications","doi":null,"url":null,"internal_url":null},{"position":23,"raw":"[Shih et al., 2018a] A. Shih, A. Choi, and A. Darwiche. For- mal verification of Bayesian network classifiers. In Proc. of PGM'18, pages 427-438, 2018.","title":"Formal verification of Bayesian network classifiers","authors":"Shih","year":"2018","venue":"Proc. of PGM'18","doi":null,"url":null,"internal_url":null},{"position":24,"raw":"[Shih et al., 2018b] A. Shih, A. Choi, and A. Darwiche. A symbolic approach to explaining bayesian network classi- fiers. In Proc. of IJCAI'18), pages 5103-5111, 2018.","title":"A symbolic approach to explaining bayesian network classifiers","authors":"Shih","year":"2018","venue":"Proc. of IJCAI'18)","doi":null,"url":null,"internal_url":"/proceedings/2018/708"},{"position":25,"raw":"[Shih et al., 2019] A. Shih, A. Choi, and A. Darwiche. Com- piling Bayesian networks into decision graphs. In Proc. of AAAI'19, pages 7966-7974, 2019.","title":"Compiling Bayesian networks into decision graphs","authors":"A Shih, A Choi, A Darwiche","year":"2019","venue":"Proc. of AAAI'19","doi":null,"url":null,"internal_url":null},{"position":26,"raw":"[Vilone and Longo, 2021] G. Vilone and L. Longo. Notions of explainability and evaluation approaches for explain- able artificial intelligence. Inf. Fusion, 76:89-106, 2021.","title":"Notions of explainability and evaluation approaches for explainable artificial intelligence","authors":"G Vilone, L Longo","year":"2021","venue":"Inf. Fusion","doi":null,"url":null,"internal_url":null},{"position":27,"raw":"[Yang et al., 2019] F. Yang, M. Du, and X. Hu. Evaluating explanation without ground truth in interpretable machine learning. CoRR, abs/1907.06831, 2019.","title":"Evaluating explanation without ground truth in interpretable machine learning","authors":"F Yang, M Du, X Hu","year":"2019","venue":null,"doi":null,"url":null,"internal_url":null},{"position":28,"raw":"[Zhou et al., 2021] J. Zhou, A.H. Gandomi, F. Chen, and A. Holzinger. Evaluating the quality of machine learning explanations: A survey on methods and metrics. Electron- ics, 10:593, 2021.","title":"Evaluating the quality of machine learning explanations: A survey on methods and metrics","authors":"J Zhou, A H Gandomi, F Chen, A Holzinger","year":"2021","venue":"Electronics","doi":null,"url":null,"internal_url":null}],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}