Event Location: NSH 1507
Bio: Shiguang Shan is now a visiting scholar with Dr. Alex Hauptmann at CMU. He received Ph.D. degree in computer science from the Institute of Computing Technology (ICT), Chinese Academy of Sciences (CAS), Beijing, China, in 2004. He then joined ICT, CAS and has been a Professor since 2010. He is now the Deputy Director of the Key Lab of Intelligent Information Processing of CAS. His research interests cover computer vision, pattern recognition, and machine learning, especially focusing on face recognition related research topics. His work on face recognition has been applied to many real-world systems in China. He has published more than 150 papers in refereed journals and proceedings in the related areas. He has served as Area Chair for many international conferences including ICCV’11, ICPR’12, ACCV’12, FG’13, ICPR’14, and ICASSP’14. He is workshop co-chair of ACCV14, and website co-chair of ICCV15. He serves as Associate Editor of IEEE Trans. on Image Processing, Neurocomputing, and EURASIP Journal of Image and Video Processing.

Abstract: Visual representation is the fundamental of many computer vision tasks. Historically, the last decade has witnessed the prosperity of local features and sparse representation. And, more recently, feature learning is blooming for learning hierarchical representations. However, most of them focus on the representation of single image or a few images. In this talk, I will introduce some of our recent works on visual representation and metrics for face classification with image sets, where multiple images are available for the class to be recognized. In our methods, the image set is collectively represented in some manifold, thus forming interesting novel problems, including distance computation between manifolds (papers in CVPR08/09/12), Manifold alignment via Universal Manifold Model(paper in CVPR14/NIPS14), and Learning Euclidean-to-Riemannian Metric (paper in CVPR2014).