Non-smooth manifold optimization with applications to machine learning and pattern recognition
Event details
| Date | 03.07.2015 |
| Hour | 10:30 |
| Speaker | Prof. Michael Bronstein, USI |
| Location |
INF119
|
| Category | Conferences - Seminars |
Numerous problems in machine learning are formulated as optimization with manifold constraints, i.e., where the variables are restricted to a smooth submanifold of the search space. For example, optimization on the Grassman manifold comes up in multi-view clustering and matrix completion; Stiefel manifolds arise in eigenvalue-, assignment-, and Procrustes problems, compressed sensing, shape correspondence, manifold learning, sensor localization, structural biology, and structure from motion recovery; manifolds of fixed-rank matrices appear in maxcut problems and sparse principal component analysis; and oblique manifolds are encountered in problems such as joint diagonalization and blind source separation.
In this talk, I will present an ADMM-like method allowing to handle non-smooth manifold-constrained optimization. Our method is generic and not limited to a specific manifold, is very simple to implement, and does not require parameter tuning. I will show examples of applications from the domains of physics, computer graphics, and machine learning.
Bio: Prof. Michael Bronstein was born in 1980. He received the B.Sc. summa cum laude from the Department of Electrical Engineering in 2002 and Ph.D. with distinction from the Department of Computer Science, Technion in 2007. In 2010, he has joined the Institute of Computational Science in the Faculty of Informatics at the University of Lugano (USI), Switzerland. Since 2012, he also serves as research scientist at the Perceptual Computing lab at Intel. He also held visiting appointments at Politecnico di Milano (2008), Stanford university (2009), INRIA (2009), Technion (2013, 2014), and the University of Verona (2010, 2014). His main research interests are theoretical and computational methods in spectral and metric geometry and their application to problems in computer vision, pattern recognition, shape analysis, computer graphics, image processing, and machine learning.
In this talk, I will present an ADMM-like method allowing to handle non-smooth manifold-constrained optimization. Our method is generic and not limited to a specific manifold, is very simple to implement, and does not require parameter tuning. I will show examples of applications from the domains of physics, computer graphics, and machine learning.
Bio: Prof. Michael Bronstein was born in 1980. He received the B.Sc. summa cum laude from the Department of Electrical Engineering in 2002 and Ph.D. with distinction from the Department of Computer Science, Technion in 2007. In 2010, he has joined the Institute of Computational Science in the Faculty of Informatics at the University of Lugano (USI), Switzerland. Since 2012, he also serves as research scientist at the Perceptual Computing lab at Intel. He also held visiting appointments at Politecnico di Milano (2008), Stanford university (2009), INRIA (2009), Technion (2013, 2014), and the University of Verona (2010, 2014). His main research interests are theoretical and computational methods in spectral and metric geometry and their application to problems in computer vision, pattern recognition, shape analysis, computer graphics, image processing, and machine learning.
Practical information
- General public
- Free
Organizer
- Signal Processing Laboratory (LTS2)
Contact
- Xavier Bresson