Séminaire ICI : Dana Lahat
Titre du séminaire et oratrice
Challenges in multimodal data fusion and multiset data analysis.
Dana Lahat, post-doctorante GIPSA-Lab
Date et lieu
Vendredi 27 juin 2014, 14h.
ENSEA, salle 384.
The last decade has been marked with a quantum leap forward in information technologies. Previously prohibitive issues such as data storage costs, data acquisition methods and data processing speed have made way to a new challenge, which is the huge amount of information that they produce. The abundance of diverse sources of information makes it practically impossible to ignore the presence of datasets that are possibly related. It is very likely that an ensemble of related datasets is "more than the sum of its parts", in the sense that it contains precious information that is lost if these relations are ignored. The information of interest that is hidden in these datasets is usually not easily accessible, however.
Despite the evident potential benefit, and significant work that has already been done in the field, the knowledge of how to actually exploit the additional diversity that multiple and heterogeneous datasets offer is currently at its very preliminary stages. In this talk, we bring together a comprehensive (but definitely not exhaustive) list of challenges in data fusion and the analysis of multiple datasets. As we illustrate in various examples, it is clear that at the appropriate level of abstraction, the same challenge can be relevant to completely different and diverse applications, goals and data types. Consequently, a solution to a challenge that is based on a sufficiently model-free approach may turn out useful in very different domains.
Our goal is to stimulate and evoke the relevance and importance of a perspective based on "challenges". More specifically, we would like to promote data-driven approaches, that is, approaches with minimal and weak priors and constraints, such as sparsity, nonnegativity, and independence among others, that can be applied to more than one specific application or dataset.
This is joint work with Christian Jutten, GIPSA-Lab and Université de Grenoble, France, and Tülay Adalı, University of Maryland Baltimore County, Baltimore, MD, USA.