You are here


  • MLG seminar - Credal classification

    On June 11, 2012 a seminar will be presented by Dr. Giorgio Corani with the title "Credal classification".  This seminar will take place on the ULB campus, building NO, Salle Rotule (8th floor) at 11:00.

  • Tom Lenaerts in ULB research news

    Act'ULB Recherche :

    L'évolution darwinienne présente la coopération (acte d'aider quelqu'un à son propre compte) comme stratégie de comportement non viable, étant donné que certains peuvent profiter directement de cet acte et que d'autres ne sont pas obligés d'agir de la même façon.

    Les biologistes théoriques ont tout de même émis un certain nombre de mécanismes expliquant l'évolution de la coopération par lasélection naturelle de Darwin. L'un des plus éminents mécanismes est laréciprocité directe de Robert Trivers qui est devenu célèbre par la stratégiedu « prêté pour un rendu ». Dans cette théorie, basée sur un jeu, les joueurs préféreraient coopérer dans une interaction réciproque s'il y a une grande chance qu'ils se rencontrent à nouveau.

    Dans l'article publié récemment dans la Physical Review LettersTom Lenaerts (Faculté des Sciences), et des chercheurs portugais, étudient comment cette idée s'applique aux groupes d'individus.  Dans ce contexte, la stratégie du « prêté pour un rendu » n'est plus applicable car l'adversaire, qui n'a pas coopéré précédemment, n'est pas connu. Une solution pourrait être de décider de coopérer si précédemment un nombre suffisant de joueurs a coopéré. Pourtant, combien de coopérateurs sont suffisants avant que d'autres décident de faire le même ?

    Tom Lenaerts et ses collègues ont montré, dans le contexte des jeux du type ressource commun, que l'évolution darwinienne mène à des stratégies qui exigent un nombre minimal de coopérateurs.  Si tous les participants tiennent à ce nombre minimal (ou plus), la coopération devient incontournable. Mais,  une stratégie moins stricte, même si cela semble plus social, donne la possibilité aux stratégies non-coopératives de profiter du système. Ces résultats sont applicables dans plusieurs domaines sociaux et scientifiques.

  • MLG Seminar - Two EMBL talks

    On March 30, 2012 Norman Davey and Kim Van Roey, both researchers at the EMBL Heidelberg, will present their research.   The presentation of Norman Davey is entitled "Understanding the role of short linear motifs in cell regulation and disease" and the presentation of Kim Van Roey is entitled "Cooperative decision-making in cell regulation".  This seminar will take place on the ULB La Plaine campus, building NO, 5th floor, Solvay room at 11:30-13:30 (STARTING TIME HAS CHANGED).    The abstract for Kim van Roey's talk can be found here.  

  • Seminar on IBM's Watson, 13/3, 12h15, UB4.228, Solbosch


    "IBM´s Watson - Deep Insight is more than just playing a game of Jeopardy !  "

    which will be presented by Marc Teerlink (IBM),
    on Tue the 13th of March from 12h15 to 14h in ULB, room UB4.228, campus du Solbosch, 1050 Bruxelles.

    How to access the conference room:


    After the resounding success of IBM´s Watson ( in the American game show Jeopardy!, the question quickly arises concerning the role of analytics and its impact on government, enterprises and consumers versus the applicability of the principles and mathematical optimization models behind Watson. Looking to those companies that have started to adopt analytics as part of their differentiation, Marc Teerlink and Michael Haydock, IBM´s Chief Scientist, observed in their recently authored study that “companies leading in use of predictive analytics and executing effectively across multiple channels have been able to increase top-line growth up to FIVE times more than their less sophisticated peer group”

    During todays session, Marc Teerlink, IBM´s Global Strategist, will take you on a roller coaster ride covering the research and facts behind IBM´s Watson winning Jeopardy!, share a framework of analytic strategies and the success they have brought to leading organizations and discuss the impact of how Deep Question and Answer (Q&A) will change society as we know it in the coming three to six years.


    Dr Marc Teerlink, mba/mbi, is a global strategist for IBM on subjects such as Business Optimization, “Business Analytics” & “Business Intelligence”. Marc is part of IBM´s Global Center of Competence, acting as a “sparring partner” and “thought leader” with IBM´s global Clients. Prior to sharing his experience with clients, for over four years Marc has been part of IBM´s internal BI & Sales Transformation team, from where his experience evolved to leading various of IBM´s geographic consulting practices.
    Marc also leads a piece of research together with leading international business schools under the title “turning data into dollars,” consolidating the lessons learned and best practices on how more than 1600 worldwide companies make money out of their customer/product related information.
    Prior to joining IBM in 2001, Marc spent 17 years as an international banker, a consultant in strategic management, a sales transformation manager in a US telecommunication firm and as an European Director for Relationship Technologies at a high-tech corporation.

    The seminar is organized jointly by Pr G. Bontempi and E. Zimanyi  in the context of the MA2 ULB course "INFO-H-508 : Questions Actuelles d'Informatique"


  • Data mining WEB interface for ICT4Rehab is online

    ICT4Rehab targets the development of assistive ICT technologies (Computer-aided gait data analysis and Kinect-based serious gaming) for the rehabilitation of children with cerebral palsy. The data mining WEB interface for ICT4Rehab is now online, and allows you to interactively test the data mining tools that are being developed in the project. Check out the interface here.

  • MLG seminar - Structure, Unstructure and Alternative Splicing

     On February 22, 2012 a seminar will be presented by Dr Philip Kim with the title "Structure, Unstructure and Alternative Splicing".  This seminar will take place on the ULB campus, building NO, Salle Sovay (5th floor in the rotule) at 16:00.   

  • MLG Seminar - Modular protein interaction domains - Evolution, Selectivity and Complexity

    On February 14, 2012 a seminar will be presented by Dr Piers D. Nash with the title "Modular protein interaction domains - Evolution, Selectivity and Complexity".  This seminar jointly organized with the MoRe group of the VIB/VUB, will take place on the VUB campus, building D, room D.005 (located at the lower level) at 11:00. A map of the VUB campus can be found here.  A barcode will be provided to those coming by car to access the VUB campus.


  • MLG Seminar - "Ensemble learning for real-world classification"

    On 18 January 2011 a seminar will be presented by Nima Hatami in the ULB (room : NO7.07) with the title "Ensemble learning for real-world classification".

  • New MLG member

    MLG welcomes Andrea Dal Pozzolo, new MLG doctoral student, which will work on the project "Adaptive real-time machine learning for credit card fraud detection" in collaboration with ATOS Worldline and funded by DOCTIRIS- INNOVIRIS.

  • PhD position available at MLG


    Applications are invited for a full-time 4-year PhD research position at the Machine Learning Group, Computer Science Department, Université Libre de Bruxelles, Belgium, starting as soon as possible.
    The researcher is expected to carry out fundamental research in (Evolutionary) Game Theory, group formation and network dynamics and to take an active role in the interdisciplinary research project in collaboration with the ULB ECARES institute, which is funding the position. Using methods of experimental economics and computational modeling we will examine the interplay group formation and preference dynamics in multiplayer games in order to provide supported answers to the origin and evolution of other regarding preferences.
    The Machine Learning Group, ULB is a research group that carries out theoretical research on statistical machine learning and learning dynamics, and takes part in several applied research projects in domains as diverse as computational biology and social dynamics.  
    Located in heart of Brussels, the Université Libre de Bruxelles is renowned for quality research through its prestigious affiliate research centres and world renowned faculty. Brussels, the capital of Belgium and the European Union, is a cosmopolitan city renowned for its picturesque squares and buildings, serene parks and gardens and a very vibrant cultural life spiced up by many international languages.


    Applicants should have a recent Master degree in Computer Science or a subject area related to statistical mechanics or computational modeling. In particular, any experience on game theory,  machine learning or artificial intelligence research will be an asset.
    Applicants are expected to have strong programming skills, good math skills, the ability to work independently, strong interpersonal skills, good writing and oral communication. 
    Upon offer of a position in MLG-ULB, the candidates who are non-EU citizens are required to obtain an appropriate long-term Schengen visa; further information is available from their nearest Belgian embassy


    Applicants are encouraged to apply as soon as possible as positions will remain open until filled by suitable candidates.


    Candidates are invited to send their CV with the detailed list of publications, if any, a research statement, a copy of their thesis abstract and the names of at least two references in .pdf format by email to  The subject field of your email must be MLG-JOB: doc.
    Candidates are also expected to attach their transcripts and a letter of recommendation from their MSc supervisor or another relevant person.

    Equal Opportunity Employer

    ULB is an equal opportunity employer and is committed to employing more handicapped individuals and especially encourages them to apply. ULB wishes to increase the proportion of women in areas in which they are underrepresented. Women are strongly encouraged to apply.


Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer