Welcome to TAO Web site


Visiting us
The Higgs boson: A Machine Learning Challenge

Who is TAO ?

TAO is the mixed INRIA Saclay CNRS LRI, Université Paris-Sud research group interested in the interplay of Machine Learning (A for Apprentissage) and Optimization (O). The two pillars of TAO's research are:
  • Machine Learning
  • Evolutionary Computation

Last activity reports: 2005 , 2006 , 2007 , 2008 , 2009 , 2010 , 2011 , 2012 , 2013 , 2014 , 2015 , 2016 (temporary version)


  • Isabelle Guyon is program co-chair of the NIPS 2016 conference.
  • Balazs Kégl, Cécile Germain and Isabelle Guyon co-organized the Higgs Boson Machine Learning Challenge , most attended ML challenge ever.
  • Best paper award in GECCO 2015 - Genetic Programming , Memetic Semantic Genetic Programming Robin Ffrancon and Marc Schoenauer
  • Best paper award in PPSN 2014 - Parallel Problem Solving from Nature , Maximum Likelihood-Based Online Adaptation of Hyper-Parameters in CMA-ES, Ilya Loshchilov, Marc Schoenauer, Michèle Sebag and Nikolaus Hansen
  • Marc Schoenauer has been elected Chair of ACM-SIGEVO, Special Interest Group for Genetic and Evolutionary Computation (2015)
  • Michele Sebag has been elected general chair of the Steering Committee of the European Association for Machine Learning and Knowledge Discovery (2015)


See also Old News

Research (under reconstruction)

The main on-going research directions/projects are:

  • Autonomic Computing - how to make a gigantic distributed computing system self-aware, self-healing, self-optimizing, self-?... (more...)
    • Project: Grid Observatory, Work Package in EGEE-III.


  • Developmental Design, Swarm Robotics and other Complex Systems. When what you optimize is the process leading to the result (this is ''embryogenesis'). more...
    • Project: SYMBRION Integrated Project, FP7.

  • Multi-Disciplinary and Continuous Optimization; dealing with computationally expensive, noisy, mixed, structured, objective functions. (more ...)
    • Projects OMD and OMD2

  • Optimal decision making under uncertainty - best illustrated by the computer-Go program MoGo? - first program to ever win over a professional Go player. (more...)

  • Reservoir Computing. How can large random structures be more efficient than carefully crafted ones (e.g. in Neural Networks) ? (more...)

Contributors to this page: sebag , evomarc , Isabelle , ggrefens , furtlehn , nicolas , brockho , auger , rros , cgp , buondia , hpaugam , alvaro.fialho , jm , monteiro , Olivier , cedric , chardon , devert and lopes .
Page last modified on Thursday 05 of October, 2017 14:20:50 CEST by sebag.