Ent?te

Research

Biologie Informatique

Leader: Hélène Touzet

La modélisation du vivant est une priorité scientifique du LIFL qui associe plusieurs équipes de recherche : Bonsai, Calcul formel, BioComputing et Dolphin.

Le champ thématique est large allant de l'analyse de séquences pour l'annotation des génomes, à la biologie systémique, la biologie synthétique en passant par le docking moléculaire et la protéomique. Ces travaux reposent sur une palette de compétences complémentaires issues de l'algorithmique discrète, de l'optimisation combinatoire, du calcul haute performance, du calcul formel, de la modélisation stochastique, des systèmes concurrents...

Cette action est par nature ouverte sur la pluridisciplinarité. Elle s'inscrit dans le PPF Bio-informatique de l'Université Lille 1, piloté par le LIFL et qui rassemble en outre 7 laboratoires de biologie avec lesquels le LIFL entretient des collaborations structurantes.

Teams

Bonsai : Algorithms for large-scale sequence analysis
Dolphin : Optimisation multi-critère parallèle coopérative
Calcul Formel : Computer Algebra
BioComputing : Programming Languages for Biological Modeling and Simulation

High Performance Computing

Leader: Nouredine Melab, Mathieu Giraud

Parallel and Distributed Computing has an important place in the research activities of LIFL. Developed in the Lab for over 20 years, this topic transverse to several research groups has recently evolved towards grid computing and more recently towards parallel computing on graphics accelerators. Favored by the installation in Lille of one site of the Grid’5000 nation-wide grid infrastructure (including GPUs) in 2005 together with one site of the EGI european-wide grid in 2010, a dynamic is created around Grid and GPU computing in connection with the Inria actions Aladdin-G5K and Hemera, France Grilles and La Maison de la Simulation. This dynamic is also favored by the organization of the potential of Université Lille 1 in HPC through the PPF « Calcul Scientifique Intensif » (2010-2013) and the international and multidisciplinary Master in Supercomputing (started in 2010).
In this context, LIFL has included in its 2010-2013 four-year plan a transverse HPC action. The objectives of such action are the following:

Conducting a collaborative thinking and discussing common challenges through feedbacks within work groups (e.g. GPGPU), through joint projects (BQR, ...), etc.
Pooling and coordinating the acquisition of HPC means for the Lab with the support of the CRI of Lille 1 (http://calcul-wiki.univ-lille1.fr).
Ensuring a scientific animation through the organization of seminars (e.g. CIGIL days) and scientific workshops, schools and specific trainings for researchers mainly PhD students.

Teams

DART : Dynamic Adaptativity and Real-Time
Shacra : Simulation in Healthcare for Advanced Medical ApplicatioNs
MINT : Méthodes et outils pour l'Interaction à gestes
Sequel : Sequential Learning
MAP : Méthodologie et Algorithmique Parallèles pour le calcul scientifique
Dolphin : Optimisation multi-critère parallèle coopérative
Bonsai : Algorithms for large-scale sequence analysis

Machine Learning

Leader: Rémi Gilleron

The machine learning research group of the computer science department is mainly composed of the members of the Sequel team and some members of the Mostrare team. The main research topics are sequential decision problems and
structured prediction. We study reinforcement learning with an emphasis on the exploration/exploitation dilemna. We define bandit algorithms and provide theoretical results. We also study classification algorithms and link prediction algorithms for graphs. The main applications are games, brain computer interfaces, recommender systems. We target to diffuse our knowledge on machine learning to other teams for 3D image, for video mining and for biology.

Teams

Mostrare : Modeling Tree Structures, Machine Learning, and Information Extraction
Sequel : Sequential Learning
Bonsai : Algorithms for large-scale sequence analysis
SMAC : Systèmes Multi-Agents et Comportements
FOX-MIIRE : Fouille et indexation de dOcuments compleXes et multimedia, Multimédia, Images, Indexation, REconnaissance

Model Driven Engineering

Leader: Laurence Duchien

The Model Driven Engineering (IDM for Ingénierie Dirigée par les Modèles) theme from the LIFL laboratory has emerged in the 90s in the GOAL team with the activities on multi-view approaches for software design. These activities have now evolved both as research objects and as tools for these objects. The ADAM, Coca, DaRT, NOCE and RMoD teams are members of this action. Several directions are explored such as adaptation, model reuse, software maintenance and modularity, along with the study of model driven engineering for application domains such as systems and HCI.

The goal of this action is enable knowledge sharing between the member teams and to help in the emergence of new research synergies. The action evolves in accordance with the directions of the Campus Intelligence Ambiante and the multi-disciplinary project that exist in the region. The action is the local representative of the IDM transversal action from the GDR CNRS GPL, ASR and I3 to which the member teams participate. A shared seminar is held periodically, the sharing of tools is promoted and some national actions are proposed by the group.

Teams

ADAM : Adaptive Distributed Applications and Middleware
Cocoa : Model Composition for Software Engineering
DART : Dynamic Adaptativity and Real-Time
NOCE : Nouveaux Outils pour La Coopération et l'Education
RMOD : Analyses and Languages Constructs for Object-Oriented Application Evolution