Section: Overall Objectives

Presentation

The last two years have been a turning point for the team. Of course, since its creation in 2003, TAO activities had constantly but slowly evolved, as old problems were being solved, and new applications arose. But recent abrupt progresses in Machine Learning (and in particular in Deep Learning) have greatly accelerated these changes also within the team. It so happened that this change of slope also coincided with some more practical changes in TAO ecosystem: following Inria 12-years rule, the team definitely ended in December 2016. The new team TAU (for TAckling the Underspecified) has been proposed, and the creation process is on-going. At the same time important staff changes took place, that also justify even sharper changes in the team focus. The year 2017 is hence the first year of a new era for the (remaining) members of the team, and some research topics might still continue to change in the next months.

As said above, several permanent members left the team. The good news is that Anne Auger and Nikolaus Hansen left TAO only to create, together with Dimo Brockoff (from the Dolphin team in Lille) their own offspring team, RANDOPT, also in the Saclay Inria Center.

The departure of both Olivier Teytaud and Yann Ollivier is a not-so-good news, in the sense that this participates to the general brain drain that is observed in all French academia toward the large private AI research centers: Olivier Teytaud (since June 2016) and Yann Ollivier (since May 2017) are now with respectively Google Zurich and Facebook AI Research in Paris. Whereas Yann continues his collaboration with us, and in particular still the supervision of his current PhD students (with more to come), Olivier's activities in TAO had to be continued by other permament researchers, namely Isabelle Guyon (see below) and Marc Schoenauer (who took over the last months of the PhD students supervisions - with Olivier in the background). We nevertheless have started a new collaboration with Olivier.

Finally, to end this presentation of the evolution of the team personal on a positive note, we were lucky enough that Isabelle Guyon joined the team (as of January 2016), as full Professor at Université Paris-Sud, and on the Inria/Université Paris-Sud Chair on Data Science for E-Science. She contributed not only by bringing her own research themes to the team, but also by enthusiastically taking over most activities of the team in the Energy application domain (see Section 4.1), the main context of Olivier's work in the recent years.

Next Sections will describe in more detail the current positionning of the new TAU team.