DIRECTIONS IN DATA SCIENCE: Michael Bronstein 26 November 2015 at 12:30 Room A14
Social Network Analysis Research Center
Data d'inizio: 26 Novembre 2015
Data di fine: 27 Novembre 2015
Multimodal spectral geometry of graphs and manifolds
Assistant Professor
Faculty of Informatics
USI Università della Svizzera italiana
- 26 November 2015
- 12:30 - 13:30
- Room A14
Spectral methods has proven to be an important and versatile tool in a wide range of problems in the fields of computer graphics, machine learning, pattern recognition, and computer vision, where many important problems boil down to constructing a Laplacian operator and finding a few of its eigenvalues and eigenfunctions (classical examples include diffusion distances, diffusion maps, and spectral clustering).
In this talk, Michael Bronstein will demonstrate how to generalize spectral geometry to settings where there are multiple data spaces. His construction of "multimodal" spectral geometry is based on the idea of simultaneous diagonalization of Laplacian operators. Next, he will show how this problem is related to the problem of finding closest commuting operators, and discuss efficient numerical methods for its solution. Bronstein uses problems from the domain of 3D shape analysis, computer vision, pattern recognition, and image processing as examples of applications.
