Electrical and Computer Engineering
Digital Signal Processing
IEEE Signal Processing Society
Director, Redwood Center for Theoretical Neuroscience
University of California at Berkeley
Highly Overcomplete Sparse Coding
Thursday, February 21, 2013
to 5:00 PM
1049 George R. Brown Hall
6100 Main St
Houston, Texas, USA
This talk explores sparse coding of natural images in the highly overcomplete regime. I show that as the overcompleteness ratio approaches 10x, new types of dictionary elements emerge beyond the classical Gabor function shape obtained from complete or only modestly overcomplete sparse coding. These more diverse dictionaries allow images to be approximated with lower L1 norm (for a fixed SNR), and the coefficients exhibit steeper decay. I also evaluate the learned dictionaries in a denoising task, showing that higher degrees of overcompleteness yield modest gains in performance. These results are of relevance to neuroscience, because the neural representation of images in cortical area V1 is also highly overcomplete. Possible advantages of overcompleteness in image representation will be discussed.
Host: Richard Baraniuk
Biography of Bruno Olshausen:
Bruno Olshausen received his B.S. and M.S. degrees in Electrical Engineering from Stanford University and a Ph.D. in Computation and Neural Systems from the California Institute of Technology (with David Van Essen). He was a postdoctoral fellow in the Department of Psychology at Cornell University (with David Field), and at the Center for Biological and Computational Learning at the Massachusetts Institute of Technology. He joined the faculty at the University of Calfornia at Davis in 1996, and in 2005 moved to UC Berkeley, where he is currently Professor of Neuroscience and Optometry. He also directs the Redwood Center for Theoretical Neuroscience, a multidisciplinary group focusing on building mathematical and computational models of brain function. Olshausen's research focuses on understanding the information processing strategies employed by the visual system for tasks such as object recognition and scene analysis. He is specifically interested in how visual representations are adapted to statistics of the natural environment.