Speaker
Xudong Jiang, Nanyang Technological University, Singapore
Topic
Data-driven Dimensionality Reduction for Pattern Recognition -- Extract Discriminative or Remove Misleading Information?
Abstract
Finding/extracting low-dimensional structures in high-dimensional data is of increasing importance, where images/signals lie in observational spaces of thousands, millions or billions of dimensions. The curse of dimensionality is in full play here: We have to conduct inference with a limited or no human knowledge. Machine learning is a solution that becomes hotter and hotter to boiling. This is evidenced by numerous techniques published in the past two decades, many of which are in prestige journals. Nevertheless, there are some fundamental concepts and issues still unclear or in paradox. For example, we often need many processing steps in a complex information discovery/recognition system. As the amount of information cannot be increased and must be reduced by any processing, why do we need them before the main processing? It is easily answerable if each step uses human knowledge. However, this seemly simple question is nontrivial in machine learning. People have proposed numerous machine learning approaches but seem either unaware of or avoiding this fundamental issue. Although extracting the most discriminative information is indisputably the ultimate objective for pattern recognition, this talk will challenge it as a proper or effective criterion for the machine learning-based dimensionality reduction, despite the fact that it has been exploited by numerous researchers. Have an in-depth thinking and express your understanding during the talk.
Speaker's biography
Xudong Jiang received the B.Sc. and M.Sc. degree from the University of Electronic Science and Technology of China, in 1983 and 1986, respectively, and received the Ph.D. degree from Helmut Schmidt University Hamburg, Germany in 1997, all in electrical and electronic engineering. From 1998 to 2004, He worked with the Institute for Infocomm Research, A*Star, Singapore, as Senior Research Fellow, Lead Scientist and appointed as the Head of Biometrics Lab. He joined Nanyang Technological University, Singapore as a faculty member in 2004 and served as the Director of the Centre for Information Security from 2005 to 2011. Currently, Dr Jiang is a tenured Associate Professor. He has published over hundred research papers in journals and conferences, where 20 papers were published in IEEE Journals: TIP(6), TPAMI(5), TSP(3), SPL(2), SPM, TIFS, TCS and TCSVT. He is also an inventor of 8 patents (3 US patents). Dr Jiang is a senior member of IEEE and elected Technical Committee Member of Information Forensic and Security, IEEE Signal Processing Society. He has been serving as Associate Editor for IEEE Signal Processing Letters and IET Biometrics. He has served as Program Committee Chair and Keynote Speaker of multiple international conferences. His research interest includes signal/image processing, pattern recognition, computer vision, machine learning and biometrics.
Invited by the Department of Electrical and Computer Engineering