报告题目 (Title):Numerical Computation for Nonnegative Matrix Factorization(非负矩阵分解的数值计算)
报告人 (Speaker): Chu Delin(新加坡国立大学)
报告时间 (Time):2025年 6 月3日 (周二) 9:00-10:00
报告地点 (Place):校本部GJ303
邀请人(Inviter):王卿文
报告摘要:Nonnegative matrix factorization is a prominent technique for data dimensionality reduction. In this talk, a framework called ARkNLS is introduced for computing NMF. first, a recursive formula for the solution of the rank-k NlS is established. This recursive form solution for the Rank-k NLS problem recursive formula can be used to derive for any integer k. As a result, each subproblem for an alternating rank-k nonnegative least squares framework can be obtained based on this closed form solution. This talk is then focused on the framework with ka new algorithm for NMFvia the closed form solution of therank3 NlS problem. Furthermore, a new strategy that efficient overcomes the potential singularity problem in rank-3 NLS within the context of NMF computation is also presented. Extensive numerical comparisons using real and synthetic datasets demonstrate that the proposed algorithm provides state-of-the-art performance in terms of computational accuracy and cpu time.