Title: Matrix Decomposition - A savoury ingredient for cooking your research
时 间： 2015年4月10日(星期五) 下午15:00-16:30
地 点： 计算机学院三楼会议室；
Dacheng Tao is Professor of Computer Science at QCIS/FEIT in UTS. He received his BEng degree from University of Science and Technology of China, his MPhil degree from the Chinese University of Hong Kong, and his PhD from the University of London. He mainly applies statistics and mathematics for data analysis problems in machine learning, data mining & engineering, computer vision, image processing, multimedia, video surveillance and neuroscience. His research results have expounded in one monograph and 300+ publications, including 100+ IEEE Transactions. He has received several Best Paper Awards/Finalists, such as the best theory/algorithm paper runner up award in IEEE ICDM’07 and the best student paper award in IEEE ICDM’13. He was/is an editor of IEEE Trans. on Image Processing, IEEE Trans. on Neural Networks and Learning Systems, IEEE Trans. on Big Data, IEEE Trans. on Knowledge and Data Engineering, IEEE Trans. on CSVT, IEEE Trans. on CYB, Pattern Recognition (Elsevier), Signal Processing (Elsevier), Information Sciences (Elsevier), Computational Statistics and Data Analysis (Springer). He is a Fellow of the IEEE, IAPR, OSA and SPIE.
Matrix Decomposition (MD) is a powerful tool in machine learning. The past decades has witnessed the effectiveness and the efficiency of MD for subsequent classification and data visualization. In this talk, we start from the simplest MD operations and the most conventional MD algorithm, Principal Component Analysis (PCA). Afterwards, we enumerate the recent progress on MD, e.g. Non-negative Matrix Factorization (NMF) and Low-Rank and Sparse Decomposition (LRSD). Then, we unfold NMF and LRSD regarding our recent developments in MD, which include GoDec and GreB, Divide & Conquer Anchoring, Regularized NMF, and Multi-view Data Modelling. Thorough experimental evidence on various real datasets and artificial datasets suggests these frameworks are effective to deal with popular research problems in computer vision and data mining.