Florida International University
Dr. Wei Zeng is as an assistant professor in the School of Computing and Information Sciences, Florida International University, and directs the Computer Graphics and Vision Lab (CGV). She received her Ph.D. in Computer Science from Chinese Academy of Sciences in 2008. She had been taking an internship in Microsoft Research Asia during 2004-2006, and visiting Stony Brook University during 2006-2009. She subsequently joined Wayne State University as a research assistant professor from 2009-2010 and then had her postdoctoral training at Stony Brook University from 2010-2012.
Dr. Zeng’s research focuses on the discrete theories, the computational algorithms, and the practical applications of modern geometry, with further interest in 3D shape analysis. Her research spans over a broad range of fields in biomedicine and engineering, including computer vision, computer graphics, visualization, medical imaging, wireless sensor network, and geometric modeling. Dr. Zeng has published 44 journal papers, 53 conference and workshop papers, 5 book chapters, and a book by Springer titled “Ricci flow for Shape Analysis and Surface Registration: Theory, Algorithm, and Application”. She has received 2 Best Paper Awards in CAD/CAM 2009 and ICCM 2017 and obtained 4 U.S. patents on geometric techniques including colon registration. She has been a technical program committee member of more than 15 conferences and has served in NSF panelist (2015) and as a frequent reviewer of Hong Kong RGC since 2013. Her research was funded by NSF.
Modern geometry has been providing powerful tools to solve real-world geometric analysis tasks. These tasks can be 3D face recognition and expression analysis, brain morphometry analysis for Alzheimer’s disease diagnosis, polyp screening in virtual colonoscopy, cardiac motion analysis, sensor routing in 3D wireless environments, and the like. The central of such tasks is to compute the shape representation of a surface and the shape metric between the surfaces, as is also the kernel of quasiconformal Teichmüller geometry theory.
This talk focuses on the theory and the computational algorithms of quasiconformal Teichmüller geometry, and their applications in practice. Teichmüller theory has its unique merits: (a) shapes in real life can be classified in Teichmüller space, because shapes with the same geometric structures are very rare; and (b) there exist a unique optimal mapping between surfaces, Teichmüller map, and a unique metric, Teichmüller metric, for shape registration and shape comparison and classification, respectively. They have offered powerful tools for shape registration, recognition, classification, and analysis, and have been widely applied in engineering and medicine. The future research goal is to explore the geometric viewpoint of machine learning based on the theories of optimal mass transport and Ricci flow and the discrete 3-manifold Ricci flow theory, for high volume geometric processing and learning tasks.