报告题目: Subspace learning: Theory and Algorithms

报告人: 张振跃教授

报告摘要: Subspace learning or segmentation aims to estimate the subspaces based on given data points from a union of several subspace. However the subspaces behind the data samples were not well defined in the literature, especially for intersected subspaces. In this talk, we build a primary frame for learning the complicated subspaces behind samples, and show the solid developments on theoretical bases and algorithms, in including the concept of fine segmentation of samples, uniqueness conditions of the fine segmentation,  structures of the subspace detection representation (SDR), a rank-restricted sparse optimization for modeling the detection of fine segmentation, tight sufficient conditions of unique solution, refinement for local optimal solutions. We may also talk about algorithms for solving the optimization problem that is neither convex nor continuous.

报告时间: 2018年12月22日 15:30--17:00

报告地点: 我院213会议室

报告人简介: 张振跃，男，浙江大学数学学院二级教授，博士生导师，浙江大学信息数学研究所所长。2013年获浙江大学心平教学杰出贡献奖，2014年获国务院政府津贴。主要从事数值代数、科学计算、机器学习和大数据分析等研究领域的模型与算法的理论分析与计算。先后在在国际著名学术刊物SIAM Review、SIAM J. Scientific Computing、SIAM J. Matrix Analysis and Application、SIAM J Numerical Analysis、 IEEE TPAMI、Patten Recognition, 以及NIPS、CVPR等会议上发表80余篇研究论文，在相关研究中取得了受到许多国际关注的系统性研究成果。他是第一位在SIAM Review上发表研究论文的国内大陆学者，其关于非线性降维算法的工作，多年来一直列SIAM J. Scientific Computing 10年高引用率第4、5位。在国际机器学习领域中被广泛应用的Scikit-Learn 中收录的8个关于流形学习的经典算法中，有两个属于其及其合作者。张振跃教授现任《计算数学》和《高校计算数学》编委。