Régis SABBADIN

De MIAT INRA
Aller à : navigation, rechercher


Régis Sabbadin has been a research scientist in the Applied Mathematics and Computer Science laboratory  in Toulouse, since 1999.

Brief bio

Diplomas:

  • "Habilitation à diriger des Recherches" from the University of Toulouse,
    on Models and algorithms for sequential decision under uncertainty in 2009.
  • PhD degree in Computer Science from the University of Toulouse,
    on An ordinal approach to decision under uncertainty: Axiomatization, logical representation and application to sequential decision-making in 1998.  
    SABBADIN2.JPG
  • Engineer degree in aeronautics from the ENSICA school (now ISAE) in Toulouse in 1993.

Research Domain

My research focuses on Artificial Intelligence methods for automated decision in sequential decision problems under uncertainty,
with applications to agricultural and natural resource management problems. The research issues are :

  • Decision theory and algorithms:
    • (sequential) decision under uncertainty,
    • structured decision problems,
    • multi-agent or multicriteria decision problem,
    • qualitative decision under uncertainty.
  • Application to the management of biodiversity, ecosystem services, pest control:
    • management of structured ecosystems: spatial ecosystems, ecological interaction networks,
    • inference of ecosystems structure: sampling optimisation and maps reconstruction, ecological networks inference,
    • biodiversity conservation, ecosystem services optimization, pests control.

The methods developed for both research issues rely, more or less heavily, on the Markov Decion Processes framework.

Selected recent publications

Around Markov decision processes

Factored and multiagent MDP
  • Sabbadin (R.) and Viet (A.F.). Leader-Follower MDP models with Factored State Space and many Followers -
    Followers Abstraction, Structured Dynamics and State Aggregation.European Conference on Artificial Intelligence (ECAI), 2016
  • Radoszycki (J,), Peyrard (N,), Sabbadin (R.). Solving F3MDPs: collaborative multiagent Markov decision processes
    with factored transitions, rewards and stochastic policies.
    Proceedings of International Conference on Principles and Practices of Multi Agent systems (PRIMA), 2015.
  • Sabbadin (R.), Peyrard (N.) and Forsell (N.). A framework and a mean-field algorithm for the local control of spatial processes.
    International Journal of Approximate Reasoning, 53(1), 66-86, 2012.
Spatial Sampling and network inference
  • Auclair (E.), Peyrard (N.), Sabbadin (R.). Labeled DBN Learning with Community Structure Knowledge.
    In: Machine Learning and Knowledge Discovery in Databases. ECML PKDD 2017. LNCS, vol 10535. Springer, 2017.
  • Bonneau (M.), Gaba (S.), Peyrard (N.) and Sabbadin (R.). Reinforcement learning-based design of sampling policies
    under cost constraints in Markov random fields: Application to weed map reconstruction.
    Computational Statistics and Data Analysis, 73, pp. 30-44, 2014.Environment and natural resources management

Applications in natural systems management

Qualitative decision under uncertainty

  • Ben Amor (N.), El Khalfi (Z.), Fargier (H.) and Sabbadin (R.). Lexicographic refinements in possibilistic decision trees and finite-horizon Markov decision processes,
    Fuzzy Sets and Systems,2018,
  • Ben Amor (N.), Fargier (H.) and Sabbadin (R.). Equilibria in Ordinal Games: A Framework based on Possibility Theory.
    Proceedings of the 26th International Joint Conference on Artificial Intelligence, pp. 105-111, 2017.
  • Ben Amor (N.), El khalfi (Z.), Fargier (H.) and Sabaddin (R.). Efficient Policies for Stationary Possibilistic Markov Decision Processes.
    Symbolic and Quantitative Approaches to Reasoning with Uncertainty. ECSQARU 2017. Lecture Notes in Computer Science, vol 10369. Springer.


For a more comprehensive list of publications.

Matlab Toolboxes

MDP Toolbox

A Matlab/Scilab/Octave/R toolbox grouping MDP solution utilities.

Authors: Chades I., Chapron G., Cros MJ., Garcia F., Sabbadin R.

 Paper: Chades I., Chapron G., Cros MJ., Garcia F., Sabbadin R. (2014). MDPtoolbox: a multi-platform toolbox to solve stochastic dynamic programming problems.
Ecography 37:916-920.

GMDP Toolbox

Authors: Cros MJ, Peyrard N., Sabbadin R.

 Paper: Cros (M.J.), Aubertot (J.N.), Peyrard (N.) and Sabbadin (R.). GMDPtoolbox: A Matlab library for designing spatial management policies.
Application to the long-term collective management of an airborne disease. PLoS ONE 12(10), 2017.

Project participations

  • French ANR project AgroBiose (2014-2017).
  • French Ecophyto project VESPA (2013-2016).
  • European project PURE (2011-2015).
  • French ANR project LARDONS (2010-2014).
  • French ANR project FICOLOFO (2010-2014).

PhD Students and Post-docs

PhD students

  • Arij Azzabi (PhD), cosupervised with N. Ben Amor (ISG, Tunis). Représentation compacte des jeux ordinaux.
    ISG-Tunis et Université Toulouse III (2018-2021).
  • Etienne Auclair (PhD), cosupervised with N. Peyrard (INRA-MIA, Toulouse).
    Apprentissage de structure d'un réseau écologique mal connu pour la conservation de la biodiversité,
    Université Toulouse III, ED MITT (2015-2018).
  • Hui Xiao (PhD), cosupervised with E. McDonald-Madden (University of Queensland) and I. Chadès (CSIRO).
    Optimal management of both ecosystem services and biodiversity, using network theory,
    University of Queensland, school of Geography, Planning and Environmental Management (2015-2018).
  • Julia Radoszycki (PhD), cosupervised with S. Gaba (INRA-SPE, Dijon) and N. Peyrard (INRA-MIA, Toulouse).
    Méthodes de conception par optimisation de stratégies spatiales de gestion de communautés de plantes adventices,
    Université Toulouse III, ED MITT(2012-2015).
  • Mathieu BONNEAU (PhD), cosupervised with S. Gaba (INRA-SPE, Dijon) and N. Peyrard (INRA-MIA, Toulouse).
    Développement d'une méthode adaptative d'échantillonnage spatialisé pour caractériser la composition du stock semencier
    des espèces adventices à l'échelle de la mosaïque paysagère
    , Université Toulouse III, ED MITT(2009-2012).
  • Nicklas FORSELL (PhD) with F. Garcia (INRA-MIA, Toulouse), L.-O. Eriksson (SLU, Umea, Sweden) and P. Wikström (SLU, Umea, Sweden).
    MDP Approaches for Complex Forest Planning Under Uncertainty, SLU, Umea (2005-2009).
  • Nasolo RAVOANJANAHARY (PhD) with J. Lang (CNRS-UPS, Toulouse). Processus décisionnels épistémiques et élicitation interactive de préférences,
    Université Toulouse III, ED MITT (2004-2008).

Post-Docs

  • Yann Dujardin, co-supervised with N. Peyrard (INRA-MIAT) and A. Vialatte (INRA-DYNAFOR).
    Multi-objective optimization applied to the management of landscapes provisioning several ecosystem services. Post-doc INRA-AgreenSkills, 2017-2019.
  • Alexandre Albore, co-supervised with N. Peyrard and F. Teichteil (ONERA, toulouse). 
    Design of mathematics and artificial intelligence tools for the mapping and sustainable management of crops pests using autonomous UAVs.
    Post-doc ONERA-INRA-Region Midi-Pyrénées, 2014-2015.
  • Will Probert, co-supervised with E. McDonald-Madden (CEED, CSIRO, University of Queensland, Australia) and N. Peyrard.
    Computational issues surrounding the dynamic optimisation of management of an ecological food web
    . Post-doc INRA-CSIRO, 2012 and 2013.
  • Alana Moore, co-supervised with M. MacCarthy (Univ. of Melbourne, Australia) and N. Peyrard.
    Control of partially observed, competitive contact processes on graphs: developing a new mathematical approach for managing invasive species or plant pathogens
    . 2010.

Teaching

I have a regular but light teaching activity (~ 15h each year). This teaching activity is linked to my research fields (operations research and decision under uncertainty).
A significant part of this teaching activity has been directed to foreign students. From 2014 on, I have decided to decrease  my teaching activities.

Recent past teaching experience :

  • MOOC Gestion Agroécologique de la Santé des Cultures et des Organismes Nuisibles (GASCON).
    3 vidéos tournées dans le grain 3 "Approches de modélisation pour la gestion agroécologique des stress biotiques:
    • "Gestion du compromis entre services écosystémiques à l'échelle du paysage : cas Cultures / Adventices / Pollinisateurs".
    • "Un cadre pour la gestion du compromis entre services écosystémiques à l'échelle du paysage: Processus Décisionnels de Markov".
    • "Modélisation du problème Culture / Adventices / Pollinisateurs dans le cadre des PDM factorisés".
  • Enseignement doctoral (ISG Tunis), Processus Décisionnels de Markov, ~20h (2017).
  • Master 2 Biostatistique et Modélisation, (UPS, Toulouse). Processus Décisionnels de Markov, 15h/year (2005-2014 et 2016).
Génotoul BioInfo
Équipe RECORD
IMABS
Outils personnels