Research

Sébastien Verel HomePage

New !

See more talks here.

top

Some of my research interests

  • Complex Systems: where some "global" properties of the system comes from a large number of "local" interactions
  • Adaptive metaheuristics: design and study methods to control the parameters of metaheuristics
  • Theory of metaheuristics, and evolutionary computation: study of the fitness landscapes of combinatorial problems specialy multiobjective and neutral ones
  • Automated Metaheuristics Design
  • Cognitive science: Design of cognitive models considering together eyes tracking and EEG signal
  • Modeling of complex systems: design and analysis of models in economics
  • Cellular Genetic Algorithm: Evolutionary algorithms where the population is structured by a grid or a graph

top

Book

"Systèmes complexes : Une introduction par la pratique"
(in french)

De Philippe Collard, Sébastien Verel et Manuel Clergue
Préface de Marco Tomassini
Publisher of EPFL Press
Presses Polytechniques et Universitaires Romandes (PPUR)
Format 160 x 240, 306 pages, mars 2013.

Retrouver sur le site web dédié, les informations concernant ce livre dont le code source.

top

Software development

  • Member of the ParadisEO development team, a software framework for the design, implementation and analysis of metaheuristics.
  • Development of the module on local search algorithms ParadisEO-MO in collaboration with J. Humeaux and A. Liefooghe.
  • Development of the tools for fitness landscapes analysis in ParadisEO-MO in collaboration with J. Humeaux and A. Liefooghe.
New version 2.0.1 of ParadisEO is available. ParadisEO is a C++ white-box object-oriented framework dedicated to the reusable design of metaheuristics: Portable (Windows, Unix and MacOS), Parallel and distributed architectures (MPI), Grids (Globus, Condor-G/MW):
  • MoCObench is a benchmark library for the Multiobjective Combinatorial Optimization.
  • MoCObench is a sourceforge project under under the GNU General Public License (GPL).
  • Instances of the Multiobjective NK-Landscapes with Correlated Objectives (rhoMNK-landscapes) are available.
  • Instance generator of the Multiobjective NK-Landscapes with Correlated Objectives (rhoMNK-landscapes) is available.
  • Multiobjective fitness functions (C, C++, java, paradiseo framework) of the Multiobjective NK-Landscapes with Correlated Objectives (rhoMNK-landscapes) are available.


Source code related to publication
  • Source code to extract the local optima network of the Quadratic Assignment Problem (QAP). Available here zip. Written in C++. Related to the works:
    • Daolio F., Tomassini M., Verel S., Ochoa G. Communities of Minima in Local Optima Networks of Combinatorial Spaces In Journal: Physica A: Statistical Mechanics and its Applications, vol. , , p. 1300 - 1308, 2011.
    • Daolio F., Verel S., Ochoa G., Tomassini M. Local Optima Networks of the Quadratic Assignment Problem In proceeding of IEEE world conference on computational intelligence (WCCI), p. 3145 - 3152, 2010-07-18, 2010.

top

National Projects

  • Member of the project (ANR) "GAZE and EEG" on the joint treatment of synchronized EEG and eye movements to monitor the spatiotemporal analysis and modeling of neuronal activities.
    Work on the modelization task with T. Baccino, B. Lemaire, and M. Clergue.
    • 2009-2011: ANR-Prog. Blanc (NT09_511856): GAZE and EEG : Traitements conjoints synchronisés de signaux EEG et de suivi de mouvements oculaires pour l'analyse spatiotemporelle et modélisation d'activités neuronales. Responsable: A. Guérin-Dugué (Université de Grenoble).


  • Project (2011) supported by the GRD Operational Research (RO) of INS2I institute (CNRS):
    • Set-based Multiobjective Combinatorial Optimization; Partners DOLPHIN team of the Université Lille 1 / INRIA, and LERIA of the université Angers, France.

top

Talks

Some of my talks from seminars and conferences:

top

Organization of events

top

Awards

  • Best paper nomination at EVOCOP (evostar) 2011:
    Verel S., Liefooghe A., Jourdan L., Dhaenens C. Pareto Local Optima of Multiobjective NK-Landscapes with Correlated Objectives, sumary and pdf.
  • Best paper nomination at GECCO 2008:
    Ochoa G., Tomassini M., Verel S., Darabos C. A Study of NK Landscapes' Basins and Local Optima Networks, sumary and pdf
  • Best paper nomination at GECCO 2006:
    Simoncini D., Verel S., Collard P., Clergue M. Anisotropic selection in cellular genetic algorithms, sumary and pdf
  • Price for best results and best paper at EvoNet Summer School, Parma, 2003.

top

Key words

artificial intelligence, operational research, evolutionary algorithms, metaheuristics, fitness landscapes, economics models, multiobjective problems, neutrality, neutral networks, combinatorial optimization, adaptive methods, complex systems, cellular automata, eyes tracking, EEG, cognitive science.

top

Some pictures

Fitness landscapes of a NK-landscapes with N=6 and K=2 with basins of attraction (see tutorial for the explanations)


Standart representations of fitness landscapes
Representation of multimodal fitness landscape


Classical representation of neutral fitness landscape


Representation by neutral networks of neutral fitness landscape

top

Program Commitees

EVO* 2013:
European Conference on Evolutionary Computation, 3-5 April, 2013, Vienna, Austria, Springer Verlag, LNCS
LION 7
Learning and Intelligent OptimizatioN, Catania - Italy, Jan 7-11, 2013, LNCS
ICANNGA'13
11th International Conference on Adaptive and Natural Computing Algorithms, April 4-6 2013, Lausanne, Switzerland, LNCS.
GECCO 2013:
Genetic and Evolutionary Computation Conference, July 06th-10th, 2013, Amsterdam (The Netherlands), ACM
PPSN 2012
11th International Conference on Parallel Problem Solving From Nature (PPSN 2012), September 1-5, 2012, Taormina, Italy.
GECCO 2012:
Genetic and Evolutionary Computation Conference, July 07th-11th, 2012, Philadelphia (USA), ACM
EVO* 2012:
European Conference on Evolutionary Computation, April 11th-13th, 2012, Malaga (Spain), Springer Verlag, LNCS
LION 5
Learning and Intelligent OptimizatioN, Rome, Italy, Jan 17-21, 2011, LNCS
EA 2011
Artificial Evolution, 24-26th October 2011, Angers, France, LNCS
GECCO 2011:
Genetic and Evolutionary Computation Conference, July 12th-16th, 2011, Dublin (Ireland), ACM
EVO* 2011:
European Conference on Evolutionary Computation, April 27th-29th, 2010, Turino (Italy), Springer Verlag, LNCS
Workshop on Self-tuning, self-configuring and self-generating search heuristics (Self* 2010)
11th International Conference on Parallel Problem Solving From Nature (PPSN 2010), September 11-15, 2010, Krakow, Poland
GECCO 2010:
Genetic and Evolutionary Computation Conference, July 7th-11th, 2010, Portland (Oregon), ACM
EVO* 2010:
European Conference on Evolutionary Computation, April 7th-9th, 2010, Istanbul (Turkey), Springer Verlag, LNCS
EVOPHD 2010:
Second European Graduate Student Workshop on Evolutionary Computation, April 7th-9th, 2010, Istanbul (Turkey)
GECCO 2009:
Genetic and Evolutionary Computation Conference, July 8th-12th, 2009, Montreal (Canada), ACM
EA 2009:
EA'09 : 9th International Conference on Artificial Evolution, October 26th-28th, 2009, Strasbourg (France), Springer Verlag, LNCS
EVO* 2009:
European Conference on Evolutionary Computation, April 15th-17th, 2009, Tubingen (Germany), Springer Verlag, LNCS
EVO* 2008:
European Conference on Evolutionary Computation, March 26th-28th, 2008, Naples (Italy), Springer Verlag, LNCS
GECCO 2007:
Genetic and Evolutionary Computation Conference, July 7th-11th, 2007, London (UK), ACM
EVO* 2007:
European Conference on Evolutionary Computation, April 11th-13th, 2007, Valencia (Spain), Springer Verlag, LNCS
EVOPHD 2007:
Second European Graduate Student Workshop on Evolutionary Computation, April 11th-13th, 2007, Valencia (Spain)
EUROGP 2006:
Nineth European Conference on Genetic Programming, April 10th-12th, 2006, Budapest (Hungary), Springer Verlag, LNCS
EVOPHD 2006:
First European Graduate Student Workshop on Evolutionary Computation, April 10th-12th, 2006, Budapest (Hungary)

top

Ph.D. thesis in computer science

Director:
Philippe Collard
Title:
"Étude et exploitation des réseaux de neutralité dans les paysages adaptatifs pour l'optimisation difficile"
pdf and sumary on HAL
slides pdf (in french)
Date
December 12, 2005
Jury :
Director :
  • M. Philippe Collard, Professor at University of Nice-Sophia Antipolis
President :
  • M. Enrico Formenti, Professor at University of Nice-Sophia Antipolis
Rapporteurs :
  • M. Cyril Fonlupt, Professor at University of Littoral - Côte d'Opale
  • M. El-Ghazali Talbi, Professor at University of Lille
  • M. Marco Tomassini, Professor at University of Lausanne
Examinateur :
  • M. Manuel Clergue, Maître de conférences at University of Nice-Sophia Antipolis
Short summary:

The works of the thesis are in the domain of combinatorial optimization by metaheuristics (evolutionary algorithms, simulated annealing, tabu search, etc). The first part consist in a theoretical work which study redudant optimization problems according to a local search in using the concept of neutral fitness landscape and neutral networks. In second part, new metaheuristics are proposed which are design for this large class of problems.



in french:

Les travaux de doctorat s'inscrivent dans le domaine de l'optimisation combinatoire par métaheuristiques (algorithmes évolutionnaires, recuit simulé, recherche tabou, etc). Ils ont consisté d'une part à l'étude des problèmes d'optimisation redondant du point de vu de la recherche locale à l'aide du concept de paysage adaptatif neutre et de réseaux de neutralité, et d'autres part à concevoir de nouvelles métaheuristiques adaptées à cette large classe de problèmes.

top

dernière modification : 13 mars 2013