Diffusion bases and their applications

Staff - Faculty of Informatics

Start date: 6 December 2010

End date: 7 December 2010

The Faculty of Informatics is pleased to announce a seminar given by Yosi Keller

DATE: Monday, December 6th, 2010
PLACE: USI Università della Svizzera italiana, SI-008,  (Via G. Buffi 13)
TIME: 15.30

The diffusion framework is a kernel-based computational approach to high dimensional data analysis and processing. It was successfully applied to high dimensional data classification and recognition. In this talk we discuss recent results on utilizing the embedding vectors as a data-driven and data-adaptive bases that can be applied to generalized domains such as tabular data sets and graphs. We start with Naoki Saito's results on the construction of bases on generalized geometrical domains and show how to apply them to sensor localization. We then present our work on image colorization and inpainting, collaborative filtering, and extension of psychometric data. Joint work with Yaniv Gur and Shai Gepstien.

Yosi Keller received the BSc degree in electrical engineering from the Technion-Israel Institute of Technology, Haifa, in 1994, and the MSc and PhD degrees in electrical engineering from Tel-Aviv University, Israel, in 1998 and 2003, respectively. From 1994 to 1998, he was an R&D officer in the Israeli Intelligence Force. From 2003 to 2006, he was a Gibbs Assistant Professor with the Department of Mathematics,Yale University. He is a senior lecturer in the Electrical Engineering Department at Bar Ilan University, Israel. His research interests include graph-based data analysis, optimization, and spectral graph theory-based dimensionality reduction.

HOST: Prof. Illia Horenko