When Sparse Learning Met Signal Processing

Stay connected



Share on facebook
Share on twitter
Share on linkedin

Vision Journal Club, May 23, 2011, Wachman 1015D

When Sparse Learning Met Signal Processing

Junzhou Huang , University of Texas, Arlington

Abstract: Sparsity techniques have been widely used to address practical problems in the fields of medical imaging, machine learning, computer vision, data mining and signal processing. The basic assumption is that the interested data/signal should be sparsely represented in some basis. However, in practical applications, the interested data/signal is not only sparse but also structured. The question is how to exploit this structured-sparsity prior information to further improve the performance of these techniques. This talk will introduce a new framework called structured sparsity, which is a natural extension of the standard sparsity techniques in statistical learning and compressive sensing. The new sparsity techniques under this framework can be successfully applied to different applications with significantly improved performance, such as compressive MR image reconstruction, video background subtraction, object tracking in visual surveillance, computer-aided diagnosis and so on. The improved experimental results demonstrate the effectiveness of this new framework.

Biography: Junzhou Huang is an assistant professor in department of computer science and engineering at University of Texas, Arlington. His major research interests are focusing on medical imaging, statistical learning and computer vision. He has won “MICCAI Young Scientist Award” in the 13th International Conference on Medical Image Computing and Computer Assisted Intervention 2010 and “FIMH Best Paper Award” in the 6th International Conference on Functional Imaging and Modeling of the Heart, 2011. He has been selected as one of the 10 emerging leaders in multimedia and signal processing by the IBM T.J. Watson Research Center in 2010.