L’IA explicable est un domaine qui a émergé avec le boom de l’apprentissage automatique et l’opacité des modèles les plus précis (comme les réseaux de neurones profonds). Les utilisateurs humains de systèmes d’IA ont, en effet, besoin de garde-fous pour éviter de tenir pour correctes des prédictions erronées et la capacité à fournir des explications des prédictions réalisées est une voie pour rejeter de telles prédictions. Ce besoin est critique lorsque des composants IA sont utilisés dans des applications sensibles et l’exigence d’explication est une affaire de réglementation en Europe (avec le RGPD depuis 2018 et avec l’AI Act maintenant). Pour ces raisons, nous avions choisi de mettre la thématique de “L’IA explicable” au centre du projet scientifique du CRIL pour le contrat 2020-2024 (étendu à 2025 suite à la pandémie COVID). En effet, il nous semblait que les compétences de l’unité sur les questions de recherche concernant les données, les connaissances et les contraintes, pouvaient être mobilisées avec profit pour créer des synergies et être à l’origine de recherches originales en IA explicable.
2024 Jérémie Bottieau, Gilles Audemard, Steve Bellart, J-M. Lagniez, P. Marquis, Nicolas Szczepanski, Jean-François Toubeau, Logic-based explanations of imbalance price forecasts using boosted trees
in Electric Power Systems Research, vol. 235, pp. 110699, 2024.
2022 Gilles Audemard, Steve Bellart, Louenas Bounia, Frédéric Koriche, Jean-Marie Lagniez, Pierre Marquis,
On the Explanatory Power of Boolean Decision Trees
in
Data and Knowledge Engineering, vol. 142, pp. 102088, 2022.
2024 Gilles Audemard, Jean-Marie Lagniez, Pierre Marquis, Nicolas Szczepanski,
Deriving Provably Correct Explanations for Decision Trees: The Impact of Domain Theories
in
The 33rd International Joint Conference on Artificial Intelligence, pp. 3688-3696, 2024.
2024 Gilles Audemard, Sylvie Coste-Marquis, Pierre Marquis, Mehdi Sabiri, Nicolas Szczepanski, Designing an XAI Interface for Tree-Based ML Models
in The 27th European Conference on Artificial Intelligence, 2024.
2024 Louenas Bounia, Frédéric Koriche,
Approximation des explications probabilistes via la minimisation super-modulaire
in
Conférence francophone sur l'Extraction et la Gestion des Connaissances EGC 2024, 2024.
2024 Gilles Audemard, Jean-Marie Lagniez, Pierre Marquis, Nicolas Szczepanski,
PyXAI: An XAI Library for Tree-Based Models
in
The 33rd International Joint Conference on Artificial Intelligence, pp. 8601-8605, 2024.
2024 Gilles Audemard, Jean-Marie Lagniez, Pierre Marquis, Nicolas Szczepanski,
On the Computation of Example-Based Abductive Explanations for Random Forests
in
The 33rd International Joint Conference on Artificial Intelligence, pp. 3679-3687, 2024.
2024 Gilles Audemard, Jean-Marie Lagniez, Pierre Marquis, On the Computation of Contrastive Explanations for Boosted Regression Trees
in The 27th European Conference on Artificial Intelligence, 2024.
2023 Gilles Audemard, Steve Bellart, Jean-Marie Lagniez, Pierre Marquis,
Computing Abductive Explanations for Boosted Regression Trees
in
Thirty-Second International Joint Conference on Artificial Intelligence {IJCAI-23}, International Joint Conferences on Artificial Intelligence Organization, pp. 3432-3441, 2023.
2023 Gilles Audemard, Jean-Marie Lagniez, Pierre Marquis, Nicolas Szczepanski,
Computing Abductive Explanations for Boosted Trees
in
26th International Conference on Artificial Intelligence and Statistics (AISTATS 2023), vol. 206, 2023.
2023 Louenas Bounia, Frederic Koriche,
Approximating Probabilistic Explanations via Supermodular Minimization (Corrected Version)
in
Uncertainty in Artificial Intelligence (UAI 2023)., vol. 216, pp. 216--225, 2023.
2023 Anthony Blomme, Daniel Le Berre, Anne Parrain, Olivier Roussel,
Compressing UNSAT Search Trees with Caching
in
ICAART 2023 : 15th International Conference on Agents and Artificial Intelligence, vol. 3, pp. 358-365, 2023.
2023 Gilles Audemard, Jean-Marie Lagniez, Pierre Marquis, Nicolas Szczepanski,
On Contrastive Explanations for Tree-Based Classifiers
in
The 26th European Conference on Artificial Intelligence (ECAI'23), IOS Press, 2023.
2023 Ryma Boumazouza, Fahima Cheikh-Alili, Bertrand Mazure, Karim Tabia,
Symbolic Explanations for Multi-Label Classification
in
15th International Conference on Agents and Artificial Intelligence (ICAART 2023), SCITEPRESS - Science and Technology Publications, vol. 3, pp. 342-349, 2023.
2023 Alexis de Colnet, Pierre Marquis,
On Translations between ML Models for XAI Purposes
in
Thirty-Second International Joint Conference on Artificial Intelligence {IJCAI-23}, International Joint Conferences on Artificial Intelligence Organization, pp. 3158-3166, 2023.
2022 Gilles Audemard, Steve Bellart, Louenas Bounia, Frédéric Koriche, Jean-Marie Lagniez, Pierre Marquis,
Trading Complexity for Sparsity in Random Forest Explanations
in
AAAI Conference on Artificial Intelligence, 2022.
2022 Gilles Audemard, Steve Bellart, Louenas Bounia, Frederic Koriche, Jean-Marie Lagniez, Pierre Marquis,
On Preferred Abductive Explanations for Decision Trees and Random Forests
in
Thirty-First International Joint Conference on Artificial Intelligence {IJCAI-22}, International Joint Conferences on Artificial Intelligence Organization, pp. 643-650, 2022.
2022 Hakim Radja, Yassine Djouadi, Karim Tabia,
Towards an FCA-Based Approach for Explaining Multi-label Classification
in
Information Processing and Management of Uncertainty in Knowledge-Based Systems - IPMU22, Springer International Publishing, vol. 1602, pp. 638-651, 2022.
2021 Ryma Boumazouza, Fahima Cheikh-Alili, Bertrand Mazure, Karim Tabia,
ASTERYX: A model-Agnostic SaT-basEd appRoach for sYmbolic and score-based eXplanations
in
CIKM '21: The 30th ACM International Conference on Information and Knowledge Management, ACM, pp. 120-129, 2021.
2021 Ryma Boumazouza, Fahima Cheikh-Alili, Bertrand Mazure, Karim Tabia,
A Model-Agnostic SAT-based Approach for Symbolic Explanation Enumeration
in
The 23rd International Conference on Artificial Intelligence (ICAI'21), pp. https://www.springer.com/series/11769, 2021.
2020 Gilles Audemard, Frédéric Koriche, Pierre Marquis,
On Tractable XAI Queries based on Compiled Representations
in
17th International Conference on Principles of Knowledge Representation and Reasoning (KR'20), 2020.
2020 Ryma Boumazouza, Fahima Cheikh-Alili, Bertrand Mazure, Karim Tabia,
A Symbolic Approach for Counterfactual Explanations
in
Davis, J., Tabia, K. (eds) Scalable Uncertainty Management (SUM2020), Springer International Publishing, vol. 12322, pp. 270-277, 2020.
2019 Karim Tabia,
Towards Explainable Multi-Label Classification
in
31st International Conference on Tools with Artificial Intelligence (ICTAI'19), 2019.