Ph.D
Group : Learning and Optimization

Uncertainties in Optimization

Starts on 01/09/2013
Advisor : TEYTAUD, Olivier

Funding :
Affiliation : Université Paris-Saclay
Laboratory : LRI-TAO

Defended on 30/09/2016, committee :

Research activities :

Abstract :
This research is motivated by the need to find out new methods to optimize a power system. In this field,
traditional management and investment methods are limited in front of highly stochastic problems which
occur when introducing renewable energies at a large scale. After introducing the various facets of power
system optimization, we discuss the continuous black-box noisy optimization problem and then some noisy
cases with extra features.

Regarding the contribution to continuous black-box noisy optimization, we are interested into finding lower
and upper bounds on the rate of convergence of various families of algorithms. We study the convergence of
comparison-based algorithms, including Evolution Strategies, in front of different strength of noise (small,
moderate and big). We also extend the convergence results in the case of value-based algorithms when dealing
with small noise. Last, we propose a selection tool to choose, between several noisy optimization algorithms,
the best one on a given problem.

For the contribution to noisy cases with additional constraints, the delicate cases, we introduce concepts from
reinforcement learning, decision theory and statistic fields. We aim to propose optimization methods closer
from the reality (in terms of modelling) and more robust. We also look for less conservative power system
reliability criteria.

CAUSAL LEARNING FOR DIAGNOSTIC SUPPORT


CAUSAL UNCERTAINTY QUANTIFICATION UNDER PARTIAL KNOWLEDGE AND LOW DATA REGIMES


MICRO VISUALIZATIONS: DESIGN AND ANALYSIS OF VISUALIZATIONS FOR SMALL DISPLAY SPACES
The topic of this habilitation is the study of very small data visualizations, micro visualizations, in display contexts that can only dedicate minimal rendering space for data representations. For several years, together with my collaborators, I have been studying human perception, interaction, and analysis with micro visualizations in multiple contexts. In this document I bring together three of my research streams related to micro visualizations: data glyphs, where my joint research focused on studying the perception of small-multiple micro visualizations, word-scale visualizations, where my joint research focused on small visualizations embedded in text-documents, and small mobile data visualizations for smartwatches or fitness trackers. I consider these types of small visualizations together under the umbrella term ``micro visualizations.'' Micro visualizations are useful in multiple visualization contexts and I have been working towards a better understanding of the complexities involved in designing and using micro visualizations. Here, I define the term micro visualization, summarize my own and other past research and design guidelines and outline several design spaces for different types of micro visualizations based on some of the work I was involved in since my PhD.