A Riemannian regularized approximate
Newton-CG method for low-rank matrix
completion
报告专家：竺筱晶副教授 （上海电力大学）
报告时间：7月17日（星期四），16：00-17：00
报告地点：数学学院西303
报告摘要：
In this talk, we propose a new Riemannian Newton-type method for the low-rank matrix completion problem. Different from the existing methods for this problem, we adopt an approximate Riemannian Hessian instead of the exact one. This approximate Hessian has simple form and several good algebraic properties such as symmetry and positive semidefiniteness. Added with a regularization term, the corresponding Riemannian Newton-type equation can be solved efficiently by the linear conjugate gradient method. Global and local convergence of the proposed method are established under mild assumptions. Preliminary numerical results on synthetic random problems of various dimensionalities and ranks are reported to demonstrate the efficiency and robustness of the proposed method.
专家简介：
竺筱晶，上海电力大学数理学院副教授，硕士生导师，数学系主任。2013年毕业于同济大学应用数学专业，2019-2020年访问日本京都大学。主要从事非线性优化方面的研究，特别是流形优化算法及理论。发表学术论文20余篇，主持国家自然科学基金项目3项，入选2024年上海市DFYC计划青年项目。
邀请人：唐庆粦
