Français Anglais
Accueil Annuaire Plan du site
Home > Research results > Dissertations & habilitations
Research results
Ph.D de

Ph.D
Group : Learning and Optimization

Elements for Learning and Optimizing Expensive Functions

Starts on 01/12/2007
Advisor : SEBAG, Michèle

Funding : Digiteo
Affiliation : Université Paris-Sud
Laboratory : LRI

Defended on 22/12/2010, committee :
* Jean-Yves Audibert (rapporteur), Ecole des Ponts-ParisTech;
* Guillaume Deffuant, Cemagref;
* Damien Ernst, Université de Liège (Belgique);
* Claudio Gentile (rapporteur), Universita' dell'Insubria, Varese (Italie);
* Michèle Sebag (directrice de thèse), Université Paris-Sud;
* Olivier Teytaud (directeur de thèse), Université Paris-Sud.

Research activities :

Abstract :
This work focuses on learning and optimizing expensive functions, that
is constructing algorithms learning to identify a concept, to
approximate a function or to find an optimum based on examples of this
concept (resp. points of the function).

The motivating application is learning and optimizing simplified models
in numerical engineering, for industrial challenges for which obtaining
examples is expensive. It is then necessary to use as few examples as
possible for learning (resp. optimizing).

The first contribution was the conception and development of a new
approach of active learning, based on reinforcement learning.
Theoretical foundations for this approach were established. Furthermore,
a learning algorithm based on this approach, BAAL, was implemented, and
used to provide experimental validation.

The approach, originally focused on machine learning, was also extended
to optimization.

The second contribution is focused on the potential and limits of both
active learning and expensive optimization, from a theoretical point of
view. Sample complexity bounds were derived: 1/ for batch active
learning; 2/ for noisy optimization.

Ph.D. dissertations & Faculty habilitations
APPRENTISSAGE ET OPTIMISATION SUR LES GRAPHES


ANALYSE DE DONNéES MULTI-MODALES POUR LES PATHOLOGIES COMPLEXES PAR LA CONCEPTION ET L’IMPLéMENTATION DE PROTOCOLES REPRODUCTIBLES ET RéUTILISABLES


DESIGNING INTERACTIVE TOOLS FOR CREATORS AND CREATIVE WORK
Creative work has been at the core of research in Human-Computer Interaction (HCI). I describe the results of a series of studies that look at how creators work, where creators include artists with years of professional practice, as well as learners, or novices and casual makers. My research focuses on three creation activities: drawing, physical modeling, and music composition. For these activities, I examine how artists switch between representations and how these representations evolve throughout their creative process, from early sketches to fine-grained forms or structured vocabularies. I present interactive systems that enrich their workflow (i) by extending their computer tools with physical user interfaces, or (ii) by making physical materials interactive. I also argue that sketch-based representations can allow for user interfaces that are more personal and less rigid. My presentation will reflect on lessons and limitations of this work and discuss challenges for future design-support tools.