Le CRIL en bref

présentation

Le Centre de Recherche en Informatique de Lens (CRIL UMR 8188) est un laboratoire de l’Université d’Artois et du CNRS qui regroupe plus de cinquante membres : chercheurs, enseignants-chercheurs, doctorants et personnels administratifs et techniques.

En savoir

Recherches en intelligence artificielle et applications

mots clés du CRIL

Actualités (RSS)

Séminaire 

Société Altifort

Séminaire Constraint-Based Symmetry Detection in General Game Playing (IJCAI 2017)

Sébastien Tabary (CRIL, Université d’Artois - CNRS)

Symmetry detection is a promising approach for reducing the search tree of games. In General Game Playing (GGP), where any game is compactly represented by a set of rules in the Game Description Language (GDL), the state-of-the-art methods for symmetry detection rely on a rule graph associated with the GDL description of the game. Though such rule-based symmetry detection methods can be applied to various tree search algorithms, they cover only a limited number of symmetries which are apparent in the GDL description. In this paper, we develop an alternative approach to symmetry detection in stochastic games that exploits constraint programming techniques. The minimax optimization problem in a GDL game is cast as a stochastic constraint satisfaction problem (SCSP), which can be viewed as a sequence of one-stage SCSPs. Minimax symmetries are inferred according to the microstructure complement of these one-stage constraint networks. Based on a theoretical analysis of this approach, we experimentally show on various games that the recent stochastic constraint solver MAC -UCB , coupled with
constraint-based symmetry detection, significantly outperforms the standard Monte Carlo Tree Search algorithms, coupled with rule-based symmetry detection. This constraint-driven approach is also validated by the excellent results obtained by our player during the last GGP competition.

En savoir

Séminaire RECAR: Recursive Explore and Check Abstraction Refinement (IJCAI 2017)

Valentin Montmirail (CRIL, Université d’Artois - CNRS)

L’approche Counter-Example Guided Abstraction Refinement (CEGAR) a été un grand succès dans la vérification de modèle. Depuis lors, elle a été appliquée à de nombreux problèmes différents. Il s’avère qu’il s’agit d’une approche pratique très efficace pour résoudre le problème QBF qui est PSPACE-complet. Dans ce talk, je vous présenterai une nouvelle approche semblable à CEGAR pour aborder des problèmes PSPACE, approche que nous appelons RECAR (Recursive Explore and Check Abstraction Refinement). Je parlerai ensuite d’une instantiation du framework RECAR pour résoudre le problème de satisfiabilité en logique modale K (problème canonique PSPACE-complet). Nous avons implémentés les deux approches CEGAR et RECAR pour déterminer la cohérence d’une formule en logique modale K au sein du solveur MoSaiC. Nous avons comparé expérimentalement ces approches face aux solveurs de l’état de l’art. L’approche RECAR surpasse l’approche CEGAR sur ces problèmes et se compare favorablement aux solveurs de l’état de l’art sur les benchmarks considérés.

En savoir

Séminaire Belief Change in a Preferential Non-monotonic Framework

Tommie Meyer (CAIR, South Africa)

Belief change and non-monotonic reasoning are usually viewed as two sides of the same coin, with results showing that one can formally be defined in terms of the other. In this talk we will discuss that it also makes sense to analyse belief change within a non-monotonic framework, and in particular we take under consideration a preferential non-monotonic framework. We consider belief change operators in a non-monotonic propositional setting with a view towards preserving consistency. We show that the results obtained can also be applied to the preservation of coherence - an important notion within the field of logic-based ontologies. We adopt the AGM approach to belief change and show that standard AGM can be adapted to a preferential non-monotonic framework, with the definition of expansion, contraction, and revision operators, and corresponding representation results.

Brief bio:
Tommie Meyer is a full professor in the Department of Computer Science at UCT, the UCT-CSIR Chair in Artificial Intelligence, and Director of the Centre for Artificial Intelligence Research (CAIR) at the CSIR in South Africa. Prior to this he was a chief scientist at the CSIR, a senior researcher at NICTA in Australia, conjoint associate professor at the University of New South Wales in Australia, associate professor at the University of Pretoria, and senior lecturer at the University of South Africa. He is recognised internationally as an expert in Knowledge Representation and Reasoning. He is a member of the steering committee of the International Conference on Principles of Knowledge Representation (KR), chair of the steering committee of the Nonmonotonic Reasoning workshop series, associate editor of the journal Artificial Intelligence, and editorial board member of the Journal of Artificial Intelligence Research. He has published in all the top journals and conferences in Artificial Intelligence.

En savoir

Prix de thèse AFIA pour Eric Piette

L’Association Française pour l’Intelligence Artificielle décerne chaque année
depuis 2009 un prix de thèse pour faire connaître et reconnaître les meilleurs
travaux de recherche des jeunes chercheurs en Intelligence Artificielle.

Cette année, la thèse d’Eric Piette intitulée “Une nouvelle approche au General Game Playing dirigée par les contraintes” reçoit ce prix ex-aequo.

Ce travail, encadré par Frédéric Koriche, Sylvain Lagrue et Sébastien Tabary
a donné lieu à la conception du joueur logiciel Woodstock, actuel champion du
monde de General Game Playing.

Eric Piette présentera son travail ce vendredi 6 juillet, à 14h, lors des Rencontres des Jeunes Chercheurs en Intelligence Artificielle (RJCIA), à la plateforme IA 2017.

En savoir