This page lists and links to various activities.

• Projects I have been, or am, involved in.
• Various responsabilities
• Participation to program committees, and/or organization of conferences
• Expertize
• Skills
• Software development

Projects

On-going academic collaborations:

• March 2019: project ``Bandits for Health'' (B4H) has been accepted by i-iste ULNE. I am co-PI of this project. Starting a collaboration with François Pattou and his group at CHR Lille, Inserm, and Université de Lille. More information through this link.
• February 2019: starting a collaboration with Romain Gautron (PhD student at CIRAD) on using sequential decision making under uncertainty for sustainable development in agriculture in developing countries.
• February 2019: Inria project Lab HY_AIAI (2019-2022) has been accepted.
• Since January 2019, collaboration with Tanguy Levent and Yvan Bonnassieux, PICM laboratory at École Polytechnique, on smartgrid control with reinforcement learning.
• Dec. 2018: Project RAID has been accepted for funding aiming at investigating the use of deep learning in radiology. Project done in collaboration with radiologists at CHRU Lille (2019-2020). PI.
• Fall 2018: starting collaborations with Christian Duriez, Defrost, on the use of reinforcement learning to control soft robots (2018-2021).
• Participation to Inria Project Lab: HPC Big-Data (2018-2021).

On-going collaborations with companies:

• Feb-Sep 2019: advising a start-up that does not want to be named
• Critéo, on computationl advertizing, through a PhD student grant (CIFRE), Oct. 2017-2020
• Renault, on the autonomous vehicle, through a PhD student grant (CIFRE), Oct. 2017-2020

On-going collaborations with PhD Students:

• Pierre Schegg, Defrost, funded by Robocath (supervised by Ch. Duriez and J. Desquidt).
• Yannis Flet-Berliac, funded by Université de Lille
• Romain Gautron, funded by CIRAD and CIAT/CGIAR (supervised by Marc Corbeels, CIRAD).
• Tanguy Levent, PICM, École Polytechnique (supervised by Y. Bonnassieux)
• Mathieu Seurin, funded by BabyRobot project (co-supervised with Olivier Pietquin)
• Kiewan Villatel, funded by Critéo (co-supervised with Jérémie Mary and Elena Smirnova, Critéo).

Responsabilities

Research management (on-going) :

• Head of SequeL, Inria team-project (2007-2018), joint group with UMR CNRS CRIStAL.
• Délégué scientifique adjoint (deputy scientific head), Inria Lille, since 2016.
• Member of the Inria commission d'évaluation (evaluation committee, CE) since 2016.
• Member of the Project Committee Board (bureau du comité des projets) of Inria Lille since 2015.
• Member of the Project Committee (comité des projets) of Inria Lille Nord Europe Research Center since 2006.

In the past:

• Head of the ``Data Intelligence'' thematic group within CRIStAL (2013-Jan. 2019).
• Member of the scientific committee of CRIStAL (2015-Jan. 2019).
• Head of the LIFL at Université de Lille 3 (2005-2007, 9/2013-12/2014).
• Founder and director of the Laboratoire d'Informatique du Littoral at Université du Littoral Côte d'Opale, Calais, France (1994-1/2003).
• Head of Alg research group at LIL (1994-1/2003).

Cursus:

• Head of the master in « machine learning & Data science » (closed)
• Head of the master Mathématiques et Informatique Appliquées aux Sciences Humaines et Sociales (MIASHS), Université de Lille 3 (2005-2007)
• Head of the DEA « Modélisation et Simulation des Systèmes Complexes » (computer science and signal processing diploma), Université du Littoral Côte d'Opale (2002-2003)

Participation to conference organization

• Organization of scientific events:
  • co-organizer of the 1st summer school entirely dedicated to RL and bandits: Reinforcement Learning Summer Scool in Villeneuve d'Ascq next July (2019).
  • co-organizer of the 14^th European Workshop on Reinforcement Learning (EWRL), Oct 1-3, Lille.
  • co-organizer of Big Data : Modelisation, Estimation and Selection, June 2016
  • Local organization chair for ICML 2015, Lille
  • PC co-chair CAP, Villeneuve d'Ascq, 3-5 juillet 2013
  • Challenge Exploration/Exploitation 2012 based on data provided by Yahoo!,
    and the accompanying workshop on new Challenges for Exploration & Exploitation.
  • co-organizer of the 8^th European Workshop on Reinforcement Learning, June-July 2008, Villeneuve d'Ascq, Post-proceedings in the Springer, LNAI series, volume 5323.
  • co-organizer of the workshop Kernel Machines and Reinforcement Learning during ICML, Pittsburgh, June 29^th, 2006
  • workshops chairman of 2006 (janvier) à Villeneuve d'Ascq
  • main organizer of the journées PDMIA, juin 2005 à Villeneuve d'Ascq (french speaking days on Markov Decision Processes and Reinforcement Learning)
  • co-organizer of the series of meetings « Systèmes complexes et SHS », Université de Lille 3, 2004-2007
  • local organization of Artificial Evolution 1999, Dunkerque
• SPC of IJCAI 2018
• PC member of:
  • International Conference on Machine Learning (ICML): 2018, 2013, 2012, 2010
  • ICML workshops: Prediction with Sequential Models workshop at ICML'2013
  • AI & Stats, 2017
  • European Conference on Machine Learning (ECML/PKDD): 2019, 2018, 2017, 2016, 2009, 2007
  • IJCAI: 2019, 2015 Machine learning track, 2009
  • 14th EWRL 2018, 10^thEuropean Workshop on Reinforcement Learning, held in conjunction with ICML'2012, in Edinburgh
  • Artificial Evolution: 2007, 2005, 2003, 2001, 1999.
  • IEEE International Symposium on Approximate Dynamic Programming and Reinforcement Learning (ADPRL): 2007, 2009, 2011, 2013, 2014, 2015, 2016
  • AIMLAI-XKDD (Joint International Workshop on Advances in Interpretable Machine Learning and Artificial Intelligence & eXplainable Knowledge Discovery in Data Mining) at ECML/PKDD 2019
  • 1st Workshop on Transfer in Reinforcement Learning (TiRL) 2017
  • Int'l Conf. on Pattern Recogntiion Applications and Methods (ICPRAM): 2012, 2013, 2014, 2015, 2016
  • Conférence Francophone sur l'Apprentissage Automatique : 2010, 2014
  • Extraction et Gestion des Connaissances (EGC: french-speaking conference on Data Mining): 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019
  • les Journées Francophones Planification, Décision, Apprentissage pour la conduite de système (JFPDA): 2006, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018
  • journées de la Société Francophone de Classification: 2015, 2016, 2017, 2018.
  • and others...
• Reviewer for:
  • ICML : 2014, 2015, 2019
  • NIPS : 2010, 2011, 2012, 2013, 2014, 2016, 2017 2018
  • AI & Stats 2017, 2018, 2019
  • ...

(some) Older projects:

• work with Sidexa, on deep learning and vision (2017-2018)
• the ANR-Lampada (2010-2014)
• the FUI Hermès project (2013-2016)
• collaboration with Squoring Technologies (2010-2014), Toulouse
• I was a member of the late Pascal network of excellence
• the Ubiquitous Virtual Seller (2009-2011), from the ``Pôle de Compétitivité Industries du Commerce'', Région Nord-Pas de Calais/FEDER funding
• the ANR EXPLORA: EXPLOration - EXPLOitation for efficient Resource Allocation. Applications to optimization, control, learning, and games. (2009-2011)
• contract with Addressing Business, Lille (2010-2011)
• contracts of research with Orange Labs
  • contract in 2009 on ad display on websites
  • contract 2009-2012 on sequential machine learning
  • contract 2014-2017 on sequential machine learning
  • a PhD student grant (CIFRE), with Dr. Tanguy Urvoy, 2014-2017
• the DYNAPP (= Learning Dynamics) project, 2004-2006
• the ABC (Behavior-based animations) Robea project, in 2003
• the ACI Cognitique project, 2000-2003
• looking for the late « Groupe Apprentissage par REnforcement » (GARE = Reinforcement Learning Group): the webpage is gone! See SequeL for my current activities in reinforcement learning.

Expertize

Beyond scientific paper reviewing for conferences and journals:

• I regularly expertize proposals for the AERES, ANR, ANRT, Inria, CNRS, among others.
• Hiring committees for assistant professor and professor at universities, and researchers at Inria.
• Member of the Inria Evaluation Commission since 2016.

Skills

I got acquainted with computer science in the early 1980's at the very exciting period of micro-computers. I learnt some of the most important things about programming with a simple TI 57 programmable calculator; then, I became proficient with programming with an Oric 1 micro-computer. Then I was trained as a computer scientist in the 1980's. Since then, I learnt by myself a whole lot of maths, and statistics.

• Programming languages:
  • high proficiency: C, R
  • decent proficiency: python, Ada, Prolog, Java
  • former high proficiency: various lisp dialects (Le_Lisp 15 for those who see what I mean), Smalltalk, Pascal, various assembly languages and machine code (6502, 8080, 8088, 8086, 68000, Z80, Cray Y-MP aka CAL)
  • older stuffs: Cobol, Pl/1, Fortran, Forth
• OS: Unix, Multics
• Compiler design, and code generation
• ...

I've been fascinated by artificial intelligence since my earliest contacts with programmable devices. Around 1981, I tried to program my TI 58 to solve the Rubik's cube, and to play Yams. Then, I discovered AI languages such as Lisp, Prolog, and Smalltalk. I implemented a subset of a Le_Lisp interpretor on my Oric-1, hand-coded in 6502 code machine (those readers who are less than 30y old can probably not even understand what I mean!). As an undergrad, I tried (and failed!) to implement a Prolog engine. My first AI related personnal work during my studies has been related to Smalltalk. Then, I did my PhD which was not AI related. Afterwards, I got interested in neural networks, and genetic algorithms. Then, I eventually touched upon machine learning around the end of the 1990's, in particular reinforcement learning, a topic that has remained my research field since then.

Software

Developing software is really a recreation for me. I am so deeply involved in management activities, and teaching activities, that coding are extra great moments for me. Unfortunately, I do not have much time for it and I do not want to release code that has not been debugged enough. So, I only advertize here two codes I have written:

• ECON: based on Manuel Loth PhD work (see the related publication). In C.
• StoSOO: which is a code for the global optimization of multi-dimensional real noisy functions. In C.

I have repeatedly coded in C ML software that had been prototyped in Matlab by collaborators. Each time, I have had the great pleasure of hearing: "Hey Philippe, your code is so much faster than mine in Matlab!".
To me, this sentence says it all. Indeed, on codes I have written, a speed-up of 1000x is not uncommon when moving from Matlab to C, plus the possibility of handling very large datasets, and streams of data.
(and I don't even mutli-thread the code, nor use GPU.)

Back to homepage.