Advanced Lecture Series in Pattern Recognition
题 目 (TITLE)：A Tensor Algebraic Framework for Image Science
讲 座 人 (SPEAKER)：Prof. Demetri (University of California, Los Angeles)
主 持 人 (CHAIR)： Prof. Liang Wang
时 间 (TIME)：September 21(Monday), 2015, 10:30AM
地 点 (VENUE)：No.1 Conference Room (3rd floor), Intelligence Building
I will present a multilinear (tensor) algebraic framework for image science, which mathematically models the fact that natural images result from the interaction of multiple factors related to scene geometry, illumination, and imaging. Within this framework, numerical multilinear algebra provides a principled approach to disentangling and explicitly representing each of these causal factors for the purposes of image synthesis, analysis, and recognition. Our learning-driven image modeling techniques employ the N-Mode SVD, a (nonlinear) tensor generalization of the conventional (linear) matrix Singular Value Decomposition (SVD). This leads to multifactor generalizations of Principal Components Analysis (PCA) and Independent Components Analysis (ICA), as well as to multifactor manifold learning methods corresponding to Isomap, Locally Linear Embedding, and Laplacian Eigenmaps. We have applied our tensor algebraic techniques to important problems in computer graphics, computer vision, and pattern recognition, among them image-based rendering, specifically the synthesis of images of natural textured surfaces under varying viewpoint and illumination, and the multilinear analysis of facial image ensembles for the automated recognition of people's faces under unconstrained view, illumination, expression, and other conditions.
Demetri Terzopoulos is Chancellor's Professor of Computer Science at UCLA, where he holds the rank of Distinguished Professor and directs the UCLA Computer Graphics & Vision Laboratory. He is or was a Guggenheim Fellow, a Fellow of the ACM, IEEE, Royal Society of London and Royal Society of Canada, and a Member of the European Academy of Sciences and Sigma Xi. Among his many awards are an Academy Award for Technical Achievement from the Academy of Motion Picture Arts and Sciences for his pioneering work on physics-based computer animation, and the inaugural Computer Vision Distinguished Researcher Award from the IEEE for his pioneering and sustained research on deformable models and their applications. One of the most highly cited authors in engineering and computer science according to ISI and other indexes, his publications include more than 300 research papers and several volumes, primarily in computer graphics, computer vision, medical imaging, computer-aided design, and artificial intelligence/life. Prior to joining UCLA in 2005, Dr. Terzopoulos held the Lucy and Henry Moses Endowed Professorship in Science at New York University and was Professor of Computer Science and Mathematics at NYU's Courant Institute of Mathematical Sciences. Previously, he was Professor of Computer Science and Professor of Electrical and Computer Engineering at the University of Toronto. He received his PhD degree in EECS from the Massachusetts Institute of Technology (MIT) in 1984. (http://www.cs.ucla.edu/~dt)