Computational methods for multi-linear problems in vision and learning



报告题目Computational methods for multi-linear problems in vision and learning

时间地点: 201761910:00数科院第四报告厅

报告人:纪辉(Ji Hui)(新加坡国立大学)


 Many important applications in computer vision and machine learning can be formulated as bi-linear problems, e.g., blind image deblurring, vision in bad weather condition, dictionary learning, and many others. These problems are challenging ill-posed non-linear inverse problems yet see its wide applications in practice. In this talk, I will represent several models and techniques that provide feasible solutions to these challenging bi-linear problems, which is built upon  several mathematical tools, including wavelet tight frames, sparse approximation, L1-norm relating regularization, and optimization methods for non-convex problems.


JI, Hui is currently an associate professor of Dept. of Mathematics at National University of Singapore. He is affiliated with Center for Wavelet, Approximation and Information Processing.  He got his Ph.D. degree in Computer Science from University of Maryland at College Park. Since 2006, he joined NUS. His research interests include Human and Computer Vision, Computational Harmonic Analysis and Mathematical Imaging. His current research is developing sparse approximation theory and algorithms to address various challenging vision and imaging problems such as blind deblurring, image completion and object tracking. For more information, please visit

邀请人: 吴春林

南开大学 | 陈省身数学研究所 | 组合数学研究中心| 统计研究院
版权所有 南开大学数学科学学院 | 地址:天津市卫津路94号 | 邮编:300071| 电话:022-23504717