Le CRIL en bref
Le Centre de Recherche en Informatique de Lens (CRIL UMR 8188) est un laboratoire de l’Université d’Artois et du CNRS dont la thématique de recherche fédératrice concerne l'intelligence artificielle et ses applications. Il regroupe plus de 50 membres : chercheurs, enseignants-chercheurs, doctorants et personnels administratifs et techniques. Le CRIL bénéficie du soutien du Ministère de l’Enseignement Supérieur et de la Recherche, du CNRS, de l’Université d’Artois, de l’IUT de Lens et de la région Hauts de France.
Recherches en intelligence artificielle et applications
Actualités (RSS)
Séminaire La logique de typicité propositionnelle
Ivan Varzinczak - CRIL CNRS & Univ Artois
Dans cet exposé, je présenterai la logique de typicité propositionnelle PTL. Cette logique enrichît la logique propositionnelle classique avec un opérateur unaire permettant de représenter de façon explicite une notion de typicité d’une formule. L’intuition derrière cet opérateur est de capturer les situations les plus typiques (ou normales, ou conventionnelles, en fonction de l’application en question) où une formule donnée est vraie. La sémantique de PTL est en termes de modèles préférentiels, plus précisément les modèles modulaires, étudiés dans l’approche KLM pour le raisonnement non monotone. Cela nous permet de montrer que les conditionnels à la KLM peuvent être simulés directement dans PTL. Je présenterai certaines propriétés intéressantes de PTL, notamment (1) le fait qu’il y a des formules de PTL qui ne peuvent pas être exprimées comme des conditionnels KLM, et (2) le fait que PTL nous permet de simuler la révision de croyances à la AGM directement dans le langage objet. Cela fait de PTL un formalisme à la fois simple et suffisamment puissant pour servir de framework dans lequel analyser des approches pour le raisonnement non monotone en logique propositionnelle. Je finirai avec une discussion à propos de différentes notions de conséquence logique non monotones dans le cadre de PTL et j’en proposerai trois candidates appropriées.
Pour plus de détails : https://arxiv.org/abs/1809.10946
Séminaire Beyond NP Revolution
**Kuldeep Meel - National University of Singapore **
The paradigmatic NP-complete problem of Boolean satisfiability (SAT) solving is a central problem in Computer Science. While the mention of SAT can be traced to early 19th century, efforts to develop practically successful SAT solvers go back to 1950s. The past 20 years have witnessed a “NP revolution” with the development of conflict-driven clause-learning (CDCL) SAT solvers. Such solvers combine a classical backtracking search with a rich set of effective heuristics. While 20 years ago SAT solvers were able to solve instances with at most a few hundred variables, modern SAT solvers solve instances with up to millions of variables in a reasonable time.
The “NP-revolution” opens up opportunities to design practical algorithms with rigorous guarantees for problems in complexity classes beyond NP by replacing a NP oracle with a SAT Solver. In this talk, we will discuss how we use NP revolution to design practical algorithms for two fundamental problems in artificial intelligence and formal methods: Constrained Counting and Sampling
Bio:
Kuldeep Meel is an Assistant Professor of Computer Science in School of Computing at the National University of Singapore where he holds the Sung Kah Kay Assistant Professorship. He received his Ph.D. (2017) and M.S. (2014) degree in Computer Science from Rice University. He holds B. Tech. (with Honors) degree (2012) in Computer Science and Engineering from Indian Institute of Technology, Bombay. His research interests lie at the intersection of Artificial Intelligence and Formal Methods. Meel has co-presented tutorials at top-tier AI conferences, IJCAI 2018, AAAI 2017, and UAI 2016. His work received the 2018 Ralph Budd Award for Best PhD Thesis in Engineering, 2014 Outstanding Masters Thesis Award from Vienna Center of Logic and Algorithms and Best Student Paper Award at CP 2015. He received the IBM Ph.D. Fellowship and the 2016-17 Lodieska Stockbridge Vaughn Fellowship for his work on constrained sampling and counting.
Séminaire Imprecise probabilities in supervised learning: principles and challenges
Sébastien Destercke
Iin this talk, we will first introduce and motivate the verybasic principles of imprecise probabilistic (IP) approaches, that
replace precise-valued probabilities and expectations by bounded
intervals. The talk will then focus on the use of IP in supervised
classification, using the extension of the naive Bayes classifier as
an illustration. Finally, we will briefly mention some challenges
concerning the application of IP to supervised learning over
combinatorial domains.
Séminaire Introduction au réseaux de neurones profonds : keep calm and do gradient descent.
Jean-Baptiste Gouray
Ces deux exposés consisteront en des travaux pratiques en jupyter notebook (python). L’idée est de comprendre sur un exemple très concret le fonctionnement et l’apprentissage d’un neurone. C’est pourquoi nous coderons la régression linéaire et l’algorithme de descente de gradient. Experts s’abstenir. N’oubliez pas d’installer jupyter sur votre portable. N’hésitez pas à nous contacter pour toute question (yael.fregier@gmail.com et jeanbaptiste.gouray@gmail.com).
Séminaire Learning Tractable Probabilistic Models in Open Worlds
Amélie Levray
Large-scale probabilistic representations, including statistical knowledge bases and graphical models, are increasingly in demand. They are built by mining massive sources of structured and unstructured data, the latter often derived from natural language processing techniques. The very nature of the enterprise makes the extracted representations probabilistic. In particular, inducing relations and facts from noisy and incomplete sources via statistical machine learning models means that the labels are either already probabilistic, or that probabilities approximate confidence. While the progress is impressive, extracted representations essentially enforce the closed-world assumption, which means that all facts in the database are accorded the corresponding probability, but all other facts have probability zero. The CWA is deeply problematic in most machine learning contexts. A principled solution is needed for representing incomplete and indeterminate knowledge in such models, imprecise probability models such as credal networks being an example. In this work, we are interested in the foundational problem of learning such open-world probabilistic models. However, since exact inference in probabilistic graphical models is intractable, the paradigm of tractable learning has emerged to learn data structures (such as arithmetic circuits) that support efficient probabilistic querying. We show here how the computational machinery underlying tractable learning and inference has to be generalised for imprecise probabilities. Our empirical evaluations demonstrate that our regime is also effective.