报告题目：Efficient adaptive randomized algorithms for fixed-threshold low-rank matrix approximation

报 告 人：刘巧华副教授（上海大学）

报告时间：2025年1月16日  11:15-12:00

报告地点：格物楼528

报告摘要：

This talk presents an adaptive randomized rank-revealing algorithm of the data matrix A, in which the orthogonal basis matrix of the approximate range space is adaptively built block by block, through a recursive deflation procedure on A.Detailed analysis of randomized projection schemes are provided to analyze the numerical rank reduce during the deflation, as well as the spectral and Frobenius error of the approximate low-rank matrix and the error of approximate singular values. The blocked randomized algorithm behaves more reliable and more efficient than the existing rank-revealing algorithm in applications arising in image processing and background estimation problems.

报告人简介：

刘巧华，博士，上海大学数学系副教授，主要研究广义最小二乘问题的扰动和条件数、矩阵逼近的随机化算法。已主持国家自然科学基金2项，上海市自然科学基金，上海市教委科研项目各1项。研究成果发表在SIAM J. Matrix Anal. Appl., SIAM J. Sci. Comput., J. Sci. Comput., Numer. Linear Algebra Appl.等国际学术期刊上。