报告题目:Multidimensional Imaging Data Completion via Weighted Three-Directional Minimax Concave Penalty Regularization
报告人:黎稳 教授 单位:华南师范大学\韩山师范学院
报告时间:2026年10月17日(周六)上午 9:30-10:30
报告地点:数学学院A321
欢迎全校师生参加!
数学学院
报告人及报告内容简介:
黎稳,1996年于西安交通大学获博士学位。现为华南师范大学二级教授、韩山师范学院“韩江学者”杰出教授、广东省计算数学学会副理事长。曾任华南师范大学数学科学学院院长、中国数学会理事、中国计算数学会理事、广东省数学副理事长、广东省工业与应用数学会副理事长。
主要研究方向:数值代数及其应用、张量计算与应用。主持国家自然科学基金面上项目6项。学术成果分别在Numer Math、SIAM 系列(SIOPT、SIMAX、SIIS)、IEEE系列 (TIP、TSP、TCYB、TNNLS)、Inverse Problems、Pattern Recognition、和中国科学(英文版)等重要学术刊物上发表。关于矩阵计算和张量计算的研究成果分别获2011年和2020年广东省自然科学奖二等奖(排名第一)。
Abstract:In this paper, we present a novel non-convex tensor completion model specfically tailored for multidimensional data. Our approach introducesa three-directional non-convex tensor rank surrogate regularized by the Minimax Concave Penalty (MCP) function. Crucially, the method processes data bysimultaneously exploiting low-rank structures across its three modal directions, with the MCP function effectively mitigating the over-penalization of large singular values-a common drawback in convex nuclear norm minimization. To address the inherent challenges ofthis non-convex optimization, we developan innovative approximate convex model that accurately captures the original formulation’s essence. We then develop a robust convex Alternating Direction Method of Multipliers (ADMM)-basedalgorithm, supported by a rigorousconvergenceguarantee,ensuringboththeoretical soundnessand practical reliability. Extensive experiments ona variety of real-world datasets demonstrate the superior performance and robustnessof the proposed method compared to state-of-the-art approaches.
