Multi-subspace Learning: Theory and Applications
- Chong You, Johns Hopkins University
- Time: 2018-04-02 15:45
- Host: CFCS
- Venue: Room 101, Courtyard No.5, Jingyuan
Abstract
Discovering structures in high-dimensional data, such as images and videos, has become an important part of scientific discovery in many disciplines. Traditional approaches often assume that the data is sampled from a single low-dimensional subspace. In practice, however, high-dimensional datasets often contain multiple classes, hence they are better modeled by a union of low-dimensional subspaces.
In this talk, I will present recent development of methods for learning multiple low-dimensional subspaces from unlabeled data. In particular, I will show how these methods can be extended to handle challenges in real applications where the data is often large scale, corrupted by noise and imbalanced across classes. I will also show theoretical advancement that provides guarantees for the correctness of the presented techniques.
Biography
Chong You received two B.S. degrees in electrical engineering and in applied mathematics, respectively in 2009, and the M.S. degree in electrical engineering from Peking University, Beijing, China, in 2012. He is currently pursuing the Ph.D. degree with the Department of Electrical and Computer Engineering, Johns Hopkins University. His research interests are statistical signal processing, machine learning and computer vision.