Projects / Collaborations / Actions

BElief change FOR better MUlti-Source Information Analysis The aim of this project is to perform an analysis of pieces of information coming from different sources. The idea is to be able to produce at the same time an assessment of the reliability of the sources and a coherent view of the world that takes into account the received pieces of information. This problem of trying to find the most plausible beliefs from different sources providing conflicting pieces of information is quite a frequent pattern.

EXPlainable artificial intelligence: a KnowlEdge CompilaTion FoundATION The EXPEKCTATION project is about explainable and robust AI. It aims to devise global model-agnostic approaches for interpretable and robust machine learning using knowledge compilation: we seek for generic pre-processing techniques capable of extracting from any black-box predictor a corresponding white-box which can be used to provide various forms of explanations and to address verification queries. In the project, we plan to focus on the post hoc interpretability issue: one will consider ML models that are not intrinsically interpretable and analyze the models once they have been trained.

Event-centric Reasoning for Interpreting everydAy NArratives Making sense of everyday narratives is a very difficult task, requiring a deep understanding of language and a broad knowledge of the world. While a “human reader” can rely on high-level reasoning to draw conclusions, current models largely lack this ability. The majority of existing approaches to language understanding, especially those based on end-to-end neural models, focus primarily on performing low-level (sentence-level) forms of reasoning to accomplish tasks.

An argumentation-based platform for e-democracy E-democracy is a form of government that allows everybody to participate in the development of laws. It has numerous benefits since it strengthens the integration of citizens in the political debate. Several online platforms exist; most of them propose to represent a debate in the form of a graph, which allows humans to better grasp the arguments and their relations. However, once the arguments are entered into the system, little or no automatic treatment is done by such platforms.

Collecting, Representing, cOmpleting, merging and Querying heterogeneous and UncertaIn waStewater and stormwater network data Public and private stakeholders of the wastewater and stormwater sectors are increasingly confronted with the analysis of massive heterogeneous data (imprecise and uncertain geographical data, digital/analogue maps, etc). Obtaining accurate and updated information on the underground wastewater networks is a cumbersome task especially in cities undergoing urban expansion. This multidisciplinary project CROQUIS, designed through four work packages, aims to create a framework where researchers from Water Sciences and Artificial Intelligence will join forces in order to offer novel solutions for representing, merging, archiving, classifying, integrating domain knowledge, repairing and querying heterogeneous data capturing the main characteristics of wastewater and stormwater networks.

EXplainable and parsimonious Preference models to get the most out of Inconsistent DAtabases Nowadays, data is increasingly becoming a commodity of tremendous value for many real-world domains. Due to the heterogeneity and the imprecision of sources, gathered information is often incomplete and inconsistent. For example, social network users can easily post their observations over some events, but unfortunately such posts are frequently far from being precise or correct. Additionally, different users may report on the same event from a different point of view, consequently creating inconsistent information in the network.

Preprocessing Information for Nontrivial Goals / Advanced Compilation of Knowledge Project involving the laboratories CRIL, GREYC (Caen), IRIT (Toulouse), and LaBRI (Bordeaux). Designing algorithms ensuring fast response times is a fundamental problem in Computer Science. Its significance is all the more salient for algorithms requiring frequent interactions with humans. Indeed, one faces this issue quite often in everyday life (e.g., when using applications on the Web or on a smartphone, short response-time guarantees are mandatory).

THeory and Evidence to Measure Influence in Social structures Joint PRC with LAMSADE (PI) and LIP6. The THEMIS project aims at producing a general ordinal theory of cooperative interaction situations and power indices for the formulation of a portfolio of social ranking solutions applied to different domains of artificial intelligence, such as decision theory, game theory, computational social choice and multi-agent systems. In particular, the project will focus on the following families of problems: 1) the axiomatic design of novel social ranking solutions accompanied with a road-map of principles guiding users to the most adapted scenario; 2) the impact of the computational difficulties of algorithms for social ranking and their vulnerability to strategic behaviour; 3) the dynamics of coalition formation and the effect of social ranking solutions on the behaviour of individuals to form stable coalition structures; 4) the application of compact preference representation to efficiently compute social ranking solutions and to assess their robustness to changes and to manipulations.

Enhancing B Language Reasoners with SAT and SMT Techniques This is a joint project with INRIA Nancy (PI), ULiège and Clearsy. The BLaSST project targets bridging combinatorial and symbolic techniques in automatic theorem proving, in particular for proof obligations generated from B models. It focuses on advancing the state of the art in automated reasoning, in particular SAT and SMT techniques, and on making these techniques available to software verification. More specifically, encoding techniques, optimized resolution techniques, model generation, and lemma suggestion will be considered.

Efficient Query answering Under UpdateS Efficient query answering, i.e., computing the answer to a query on a given database, is one of the core problems studied in database theory. It is a very fruitful area of research with a long history and many new results and directions, e.g. efficient algorithms for aggregation, enumeration of query answers, and provenance computation. Although in practice databases are dynamic objects changing over time, the theoretical research on this topic has largely focused on static databases: when the database changes even slightly, the algorithms have to be rerun from scratch before answering, losing all already computed information.

Partners : UCCS, LML, LEM, CDEP Within the ANR project (Doctoral contracts in artificial intelligence - establishment), the programme strategy of the University of Artois is to promote and conduct core research on explainable AI as well as quality interdisciplinary research on important issues in Social Sciences and Chemistry. Five PhD contracts have been supported by the ANR, and two additional ones by the University of Artois.

ARgumentation tHeory And natural language ProceSSing fOr e-DemocracY Joint project with Prof. Mihai Surdeanu from the University of Arizona. The RHAPSSODY project proposes an automated reasoning framework that can extract, understand, and reason with complex arguments. The project is based on combining computational argumentation theory with natural language processing. The goal is to identify the most important arguments listed on debate platforms, estimate the acceptability degrees of these arguments using information mined from the web, and, using the totality of the arguments (those from the particular debate and those mined from the other web-sites), estimate the decision that will be taken.

Modern Approaches to Knowledge Compilation MAKC is an international research project (IRP) shared between the Automated Reasoning Group of the University of California at Los Angeles (UCLA) and the Centre de Recherche en Informatique de Lens (CRIL UMR 8188 CNRS - Artois University). It has been created in 2020 for five years and is funded by CNRS, the University of California and Artois University. The MAKC IRP is centered on knowledge compilation (KC) for problem solving.

Partners: INRIA, Université de Lille, IMT Douai, CNRS, UPHF, ULCO, UPJV The CornelIA project of the Hauts-de-France CPER (State-Region Plan Contract) aims to lay the foundations for responsible and sustainable AI, with scientific responses to this major challenge, a multidisciplinary work of co-construction and a in a progressive situation. It has four axes: the theoretical and scientific bases of AI, embedded AI and societal issues, multidisciplinary links and applications, and the socio-economic impact, mediation and the creation of a center of competence in AI.

Foundations of Trustworthy AI – Integrating Reasoning, Learning and Optimization TAILOR is an H2020 European project dedicated to trustworthy AI and funded by ICT-48-2020 ``Towards a vibrant European network of AI excellence centres’’ call. TAILOR brings together 55 European partners (industrial and academic). The purpose of TAILOR is to build the capacity of providing the scientific foundations for Trustworthy AI in Europe by developing a network of research excellence centres leveraging and combining learning, optimisation and reasoning.

STormwAteR and WastewAteR networkS heterogeneous data AI-driven management Acknowledgment: This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the MSCA (Marie Skłodowska-Curie Actions, Staff Exchanges)-SE (Staff Exchanges) grant agreement No 101086252; Project title: STARWARS (STormwAteR and WastewAteR networkS heterogeneous data AI-driven management). Public and private stakeholders of the wastewater and stormwater sectors are increasingly faced with large quantities and multiple sources of information/data of different nature: databases of factual data, geographical data, various types of images, digital and analogue maps, intervention reports, incomplete and imprecise data (on locations and the geometric features of networks), evolving and conflicting data (from different eras and sources), etc.

Partenaires : ULCO, UPJV, CNRS With a budget of 11 million euros over 10 years, the MAIA Project aims to support and deepen the new uses that have appeared in recent years in many scientific fields following the rise of artificial intelligence. The project aims to study, develop and deploy the strong interactions existing between artificial intelligence and three key application areas of the A2U alliance: health (UPJV), chemistry (materials, energy; UPJV/UArtois) and environment/sea (ULCO) as well as economic, sociological, ethical and legal aspects.

Produire des connaissances nouvelles en matière de justice juvénile et de santé mentale Le projet ERA vise à combler deux lacunes majeures dans les politiques publiques en faveur de la jeunesse au Sénégal : la justice juvénile et la santé mentale des jeunes. Toutes ces recherches sur la justice juvénile et la santé mentale des jeunes seront réalisées par une équipe pluridisciplinaire, composée de chercheurs (CNRS, IRD, ISED), de praticiens du droit et de la santé, d’une organisation de la société civile, l’Association Pour le Sourire d’un Enfant (APSE) et des acteurs du Ministère de la Justice du Sénégal.

Hyper-lieux, Crises, Migrations et Inégalités Le projet Hyper-lieux, Crises, Migrations et Inégalités (HYCI) propose d’analyser, au regard de deux crises majeures, de l’asile (2015-2016) et de la Covid-19, auxquelles s’ajoute la crise ukrainienne, les interactions entre les migrations internationales et les inégalités qui leurs sont attachées. HYCI s’intéresse à deux registres d’inégalités, inégalités spatiales-inégalités des droits ainsi qu’à leurs effets sur les migrant-e-s et les sociétés d’accueil, transitoire ou durable. Ces crises, passées ou en cours, révèlent la transformation et de la dynamique des parcours migratoires, interrogée dans les hyper-lieux.

Cryptanalyse algébrique pour la cryptographie post-quantum Projet en collaboration avec le MIS (porteur du projet) et le LIP6. Parmi les techniques bien établies en cryptanalyse, les attaques algébriques sont des méthodes permettant de décrire le schéma comme un système d’équations polynomiales, et par conséquent de réduire sa sécurité à la difficulté de résoudre le système associé. POSTCRYPTUM vise à concevoir des attaques algébriques efficaces, pour plusieurs classes de cryptosystèmes qui peuvent être modélisés comme un système polynomial binaire.

Production de schémas décisionnels pour atteindre l’efficience économique, sociétale et environnementale en matière de management de parcs de bâtiments et des infrastructures associées, en milieux urbains et péri-urbains.

exPlications INTElligibles État de l’art Le projet proposé, baptisé pinte, concerne le thème « caractérisation et qualification de l’IA de confiance » et plus précisément les « méthodes et la métrologie de l’explicabilité ». Il s’inscrit dans un courant de recherche pour l’IA explicable, qui se développe en France et au niveau international, et dont l’origine repose sur un double constat : à tout prédicteur (en particulier, ceux de type « boîte noire » comme les réseaux de neurones ou les forêts aléatoires) on peut toujours associer un circuit (booléen ou arithmétique), une « boîte blanche » (ou « transparente ») qui a le même comportement que le prédicteur en terme d’entrées-sorties, tout en étant intrinsèquement plus simple, les tâches d’explication (possiblement contrefactuelles) des prédictions réalisées et de vérification du fonctionnement de la « boîte noire » peuvent être réalisées en s’appuyant sur la « boîte blanche » associée.

Contrat de collaboration avec Exakis-Nelite et Aéroport De Paris. Dans le cadre des travaux d’optimisation interne du Groupe Aéroport De Paris, il a été lancé une étude autour de la mise en place d’outils de prédiction et d’optimisation des ressources aéroportuaires, pour répondre à la forte fluctuation du trafic aérien. L’objectif principal de la collaboration est d’accompagner le Groupe ADP, avec le support d’Exakis Nelite, dans la mise en place d’un outil d’aide à la décision permettant de répondre aux exigences de dimensionnement macroscopique des ressources aéroportuaire conformément aux objectifs de performance du groupe.

High Dimensional Heterogeneous Data based Animation Techniques for Southeast Asian Intangible Cultural Heritage Digital Content. “Marie Skłodowska-Curie Actions RISE (Research and Innovation Staff Exchange)” project, H2020-MSCA-RISE-2015 call. The project AniAge (High Dimensional Heterogeneous Data based Animation Techniques for Intangible Cultural Heritage Southeast Asian Digital Content) is a multidisciplinary project involving researchers in Artificial Intelligence and in computer animation. The project covers three main research topics: i) developing new digital animation techniques, ii) management of large and heterogeneous data and iii) an application to the intangible cultural heritage (IHC) of Southeast Asia countries.

Advanced Multilateral Argumentation for DEliberation The AMANDE (Advanced Multilateral Argumentation for DEliberation) project aims at providing formal tools for designing multiparty deliberation systems. The main objective is to define and study the properties of such systems, so that designers can ensure that users and possibly interacting agents are able to identify key arguments in a deliberation process, and to fairly defend their view. While the technological tools are now available to deploy such systems, we believe there is now an urgent need to work on the theoretical foundations of such platforms.

Techniques ASP pour l’interrogation d’informations web hétérogènes multisources à grande échelle. Project involving the CRIL, LERIA (Angers), LIRMM (Montpellier) and LSIS (Marseille) laboratories. The objective of the project is to propose formal methods for querying heterogeneous multi-source information at large scale. In particular, the project consists in proposing ex- tensions of the standard ASP to represent processable sub-languages of OWL2 and to study new operations of information fusion represented by logic programs in extended ASP or by OWL2-like languages of various reliability and quality with or without uncertainty.

Business Recommendation for Configurable Products. Project involving the following laboratories: CRIL (Lens), IRIT (Toulouse), LIRMM (Montpellier) and the industrialists Cameleon Software, IBM and Renault. The objective of the project is to propose formal models and algorithmic tools for online configuration applications to take into account customer preferences and implement recommendations on structured and highly combinatorial domains. See online : Project website

Interdisciplinary project on computer science and economics, involving the laboratories CREME (Caen), CRIL, LAMSADE (Paris IX) and PREG (Polytechnique). 2010-2013. MSOC is a multi-disciplinary ANR gathering researchers in computer science and economics, around the issue of computational social choice. The idea is to look at the problems of social choice (in particular voting methods) from a “computational” angle, i.e. by asking questions about representation, algorithmic complexity, etc. For example, although it is known that in theory all voting methods are manipulatable, it is possible to find voting methods whose manipulation complexity is high enough that in practice it is “impossible” (difficult) to manipulate them.

Approches déclaratives pour l’énumération de motifs intéressants. This project aims at cross-fertilization between three research areas: artificial intelligence, combinatorial algorithmics and data mining. In this context, one of the objectives is to define high-level declarative languages (logical or algebraic) to express and represent interesting pattern enumeration problems. This multi-disciplinary project gathering CRIL, LIRIS and LIMOS has allowed the emergence of a new research opening at CRIL around data mining. Many results have been obtained in this framework (see Results section).

Représentation, élicitation et agrégation de préférences sur des domaines combinatoires : nouvelles méthodes et applications. Project involving CRIL, ILOG (Valbonne), IRIT (Toulouse), LAMSADE (Paris IX), LIP6 (Paris VI) and ONERA-CERT (Toulouse). 2005-2009. The PHAC project focused on preference representation and exploitation; more specifically, it focused on the evaluation of preference representation languages, preference elicitation, preference aggregation and social choice, as well as robust, multi-criteria and uncertain decision making.

Probabilistic graphical models and description logics for alarm correlation in intrusion detection. This project is in the context of intrusion detection. It consists in providing probes with the ability to describe the observed events and to be able to reason about the alerts (correlation), while taking into account the elements of uncertainty. In particular, the project has studied new formal models of alert correlation based on the formalism of Bayesian networks and preference logics.

SAT as a Service Project involving the laboratories CRIL, CRIStAL (Lille), INRA (Toulouse) and LaBri (Bordeaux, principal investigator). The SATAS project is an ambitious project, which aims to advance the state of the art in massively parallel SAT solving with a particular eye to the applications driving progress in the field. The final goal of the project is to be able to provide a “pay as you go” interface to SAT solving services, with a particular focus on its power consumption.

Polynomiality for understanding and extending the limits of efficient solvers. The partners of this project are IRIT, LSIS, Greyc and CRIL. The TUPLES project focuses on the effective solution of NP-complete problems, in particular in the field of artificial intelligence. The main objective is to significantly push back the limits currently observed on the efficiency of combinatorial problem solvers, while at the same time establishing a theoretical framework that seems necessary to reach this objective.

This is a contract to accompany the CIFRE thesis of Benoît Trouvilliez with the company Onyme. This contract concerns the use of textual data similarities for opinion mining and product search. In particular, Benoît Trouvilliez developed a processing chain, including various NLP and learning algorithms (supervised and unsupervised) for opinion analysis from short texts. The contract was carried out in two phases: feasibility study of the clustering phase (2009) and automatic analysis and clustering of short texts for statistical purposes (2009-2012).

Advanced data science and technologies Partners: INRIA, Université de Lille 1, Mines Douai, CNRS The DATA (Advanced data science and technologies) project of the Hauts-de-France CPER (State-Region Plan Contract) is carrying out a research program on the challenges of data and associated key digital technologies (cloud computing, big data and intensive computing) in strong synergy with the regional economic fabric. It revolves around three areas of research: Internet of Things, data and knowledge intelligence, high-performance computing and optimization.

Bandits as an AUTOnomous Machine (BAUTOM) Constraint programming has developed a wide range of effective, general-purpose methods to solve many large-scale, real-world problems that can be expressed and solved as constraint satisfaction problems (CSPs). Fields that benefit from CP technology are transportation, network management, telecommunications, bioinformatics and finance [1]. As a result, organisations and industries worldwide already exploit CP technology to solve difficult problems in design and configuration, planning and scheduling, temporal and spatial reasoning, etc.

Reconciling Description Logic and Non-Monotonic Reasoning in the Legal Domain This project seeks to align Description Logics with non-monotonic formalisms in general and with the so-called Default or Defeasible Logics in particular, so that one can model dynamic situations, exceptions and other non-monotonic aspects of reasoning, such as legal reasoning and legal services argumentation. In particular, we shall investigate the barriers Description Logics are faced with; for instance, handling exceptions not explicitly contemplated in the normative literature.

Merging distributed sOurces for enhanceD quERy aNswering MODERN is a project proposal that crossovers both symbolic and numerical AI and aims to deliver robust merging methods for query-answering that are able to aggregate several and open web sources in order to get back trustful answers. In particular, the project aims to propose methods for merging that benefit from recent work on commonsense reasoning and vector space representations of entitles as tools to deal with inaccuracies encountered when aggregating information sources.

Compiling Provenance Data (CODA) PEPS JCJC INS2I 2017 impliquant le CRIL et CRIStAL UMR 9189. In this project, we will strengthen recently found links between databases and artificial intelligence, more specifically between query evaluation and knowledge compilation. The idea is that so-called provenance data that plays an important role in complex database queries can be transformed into representations that have been studied before by the artificial intelligence community. These representations then allow for tractable reasoning on the data to perform tasks from the area of databases.

Argument Patterns in computer supported cOllaborative LearNIng Computer supported collaborative learning is a pedagogical approach where learning takes place via social interaction using a computer. Previous research gives grounds to believe that groups of students that are more efficient than others use specific argument typologies in their interaction during the problem solving. Our project aims at collecting the interaction data and applying the existing approaches from computer science that allow to identify argument patterns.

Logics for Reasoning about Multi-Agent Environments Multi-agent environments are inhabited by various ‘autonomous agents’. These are are entities capable of acting autonomously, i.e., without external intervention, in order to meet their design objectives. Although single-agent environments have been studied for a long time, studies about multi-agent environments are relatively new. This kind of environment brings new issues to be dealt with, such as non-deterministic actions, competition and cooperation among agents. Current solutions treat such issues by using some kind of multi-modal logic.

The Pajero project brings together industrial (Horizontal Software, EQUITIME, CAPS) and academic (CRIL, I3S, PRISM) partners and aims to implement a multiple resource management solution capable of handling large-scale problems using innovative constraint programming techniques and parallelism, particularly in the cloud. The scientific responsibility of the project lies with CRIL. This 4-year project (2011-2015) is financed by OSEO within the ISI program (Industrial Strategic Innovation). The budget allocated to the laboratory (more than 800 k€) has allowed the recruitment of 3 PhD students, 1 research engineer (3 years) and 1 post-doctoral student.

Knowledge compilation for constraint programming The principal aim of this joint project is to use the complementary strengths of both research teams to derive new results in the research area which lies on the borderline of knowledge compilation, SAT encoding and solving, and constraint programming. In this project, the focus will be laid on the CNF format, which is the main input format for existing solvers and compilers. Our objective is to identify and study a number of CNF fragments which can serve as useful output target languages for knowledge compilation.

Méthodes de recherche locales pour l’insatisfaisabilité En partenariat avec les laboratoires MIS (Amiens), LRI (Paris Sud) et LSIS (Marseille), cette ANR a permis de proposer de nouvelles méthodes incomplètes pour l’insatisfaisabilité. De nombreux résultats ont été obtenus au cours de cette ANR : méthodes incomplètes, méthodes hybrides, résolution étendue ou encore certains résultats autour du démonstrateur GLUCOSE.

Objets Connectes Intelligents et Personnalises dans le domaine de la Nutrition (OCIP-Nut) Projet impliquant le CRIL et le LIMICS (INSERM, UMRS 1142). Le but de ce projet est de réaliser des études et des recherches préliminaires dans le but d’identifier les verrous scientifiques et cibler des pistes réalisables et à fortes retombées concernant les objets connectés intelligents dans le domaine de la nutrition. Pour ce faire, il est nécessaire d’explorer plusieurs problématiques.

Collaboration CRIL – RATP La RATP assure l’exploitation d’une partie des transports en commun de Paris et de sa proche banlieue. Dans le cadre de son activité, l’atelier de qualification logicielle (AQL) de la RATP utilise, entre autres techniques, la preuve formelle pour l’évaluation des systèmes de sécurité ferroviaire et notamment leur partie logicielle. La preuve formelle permet d’explorer l’exhaustivité des comportements d’un système et, ainsi, d’améliorer significativement le niveau de confiance dans la sécurité.

Projet commun avec la société Audace et le learning center de la Halle au Sucre de Dunkerque. Ce projet a été soutenu par Pictanovo. L’objectif global de ce projet était l’installation d’un dispositif innovant de recherche documentaire par immersion via la 3D holographique permettant de pénétrer les différentes couches et facettes du contenu documentaire d’un learning center. Nous avons développé un programme accompagnant la recherche de l’utilisateur et lui suggérant des ressources par interprétation des goûts et préférences de manière disruptive, c’est-à-dire en proposant des possibilités en lien avec les recherches actuelles mais permettant de « visiter » d’autres parties de l’espace documentaire.

sécurité des jeux en ligne Projet émergent région NPdC. L’objectif majeur du projet In Ludo Securitas (I.L.S.) est de fournir une approche automatique permettant de détecter des actes de tricherie dans le jeu en ligne mais également des comportements addictifs. Pour cela, il est absolument nécessaire de créer des outils permettant de modéliser le comportement normal d’un joueur en fonction de son niveau et de sa façon de jouer. Il semble intéressant de combiner ces informations à des données réseaux et à d’autres informations statistiques (comme le jeu des autres joueurs).

Lauréat de l’appel à projet « expériences interactives » en collaboration avec la société lensoise Audace. Le but de cette collaboration est de proposer à l’utilisateur, dans le cadre d’un jeu sérieux basé sur la métrologie, des problèmes qui s’adaptent à lui, en vue d’un comportement non déterministe de l’application, mais également pour que le jeu s’« attaque » aux faiblesses du joueur et le fasse ainsi progresser.

Le projet « Karma » est un projet de collaboration avec la société Netice, société basée à Villeneuve d’Ascq spécialisée dans les e-Porte-folio et l’usage des technolo- gies numériques pour l’emploi et la formation. Il a été amorcé par OSEO via une Prestation Technologique Réseau (PTR) et a vocation à se prolonger. Le but de ce projet est d’offrir aux formateurs et apprenants d’une FOAD (Formation Ouverte À Distance) un outil d’évaluation for- mative.

Inférence MAP dans les réseaux Bayésiens à intervalles : Approche possibiliste et application à la reconnaissance des formes. Le projet MAPPOS propose une approche originale pour le problème de l’inférence MAP dans les réseaux Bayésiens à intervalles, avec une application en reconnaissance des formes. Les requêtes MAP revêtent une importance particulière puisqu’elles sont à la base de plusieurs applications comme la classification supervisée, le diagnostic, etc. Ce problème est d’autant plus important qu’il n’existe pas d’algorithmes efficaces pour répondre à ces requêtes.

Deep Learning 4 Deep Solving Parmi les grands défis de l’intelligence artificielle figurent les problèmes « Beyond NP » apparaissant dans de nombreuses applications pratiques, telles que la planification, l’ordonnancement, la configuration et la recommandation. Ces problèmes calculatoirement difficiles, peuvent souvent être décomposés en tâches d’inférence plus simples, utilisant des tests de cohérence globale, ou d’optimisation linéaire sous contraintes. L’objectif de ce projet consiste à apprendre ces tâches d’inférence, par des réseaux de neurones profonds, qui offrent des délais de réponse très rapides, au prix d’erreurs possibles de prédiction.

Evolution des politiques de contrôle d’accès et gestion d’actions interférentes PEPS INS2I 2016 Sécurité Informatique et des Systèmes Cyber-physiques impliquant le CRIL et le LORIA CNRS UMR 7503. le contexte général de ce projet est la gestion des politiques de sécurité et leur mise en œuvre. L’observation initiale est que, dans les modèles de contrôles d’accès existants, les politiques de sécurité sont définies en termes d’actions élémentaires indépendantes du système d’information à sécuriser.

DIStributed COmpilation (DISCO) Le projet DISCO porte sur la compilation distribuée de connaissances. La compilation de connaissances vise à améliorer l’efficacité de la résolution de divers problèmes calculatoirement difficiles en réalisant un pré- traitement d’une partie des informations disponibles et en exploitant ensuite les représentations compilées calculées lors du pré-traitement. L’objectif du projet DISCO est d’étudier comment le parallélisme peut être exploité pour améliorer l’efficacité de la phase de compilation et celle d’utilisation des formes compilées produites.

fouille de données par contraintes Le projet est réalisé en collaboration avec l’université de Setif.

Non-Classical Reasoning for Enhanced Ontology-based Semantic Technologies Collaboration avec l’université de Cardiff. La tendance actuelle vers des applications intelligentes pour l’exploitation de grands ensembles de données pousse pour une meilleure compréhension des fondements théoriques et pratiques du raisonnement sur les ontologies formelles. Malgré les progrès réalisés au cours des dernières décennies dans le développement des logiques de description, de nombreux aspects importants du raisonnement humain dans ce cadre ont été largement négligés ou partiellement traités.

Etude et développement d’un système spatio-temporel de gestion de secours médicaux d’urgence Une situation d’exception ou une catastrophe est définie comme étant une grave perturbation d’une société, causant des pertes en vies humaines et en biens au point que la société sinistrée ne peut y faire face en recourant à ses seules ressources disponibles. Les catastrophes engendrent un nombre important de victimes de gravité variable. Face à de telles situations, les différents acteurs impliqués sont appelés à mettre en oeuvre, dans les plus brefs délais, les moyens indispensables permettant une prise en charge d’un afflux massif de victimes.

Qualité des Données multi-Sources, un double défi pour les sciences Sociales et les sciences de l’Informatique Ce projet questionne la qualité des données dont les modes d’acquisition sont variés et présentent différentes imperfections. Cette étude s’efforcera de développer des prétraitements, des réparations et des extractions de connaissances afin de mieux les analyser et les valoriser pour la prise de décision. Les données utilisées traitent des migrations internationales. Par la création d’outils efficaces, le projet vise à améliorer la connaissance des chercheurs et des décideurs sur l’implication des réseaux de traite dans les parcours migratoires, les relations entre le droit des migrants et les politiques migratoires, les enjeux sociaux et politiques de la mobilité des mineur(e)s.