2020/03/09 by Masoud Ahookhosh, Le Thi Khanh Hien, Ahookhosh, Masoud +5 · 1 citation
Computer Science · Engineering · Mathematics · #FOS: Mathematics #Mathematical Inequalities and Applications #Numerical Analysis (math.NA) #Optimization and Control (math.OC) #Optimization and Variational Analysis #Sparse and Compressive Sensing Techniques
paper · pdf · doi:10.48550/arxiv.2003.03963
openalex publication_date 2020/03/09 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
We propose BIBPA, a block inertial Bregman proximal algorithm for minimizing\nthe sum of a block relatively smooth function (that is, relatively smooth\nconcerning each block) and block separable nonsmooth nonconvex functions. We\nprove that the sequence generated by BIBPA subsequentially converges to\ncritical points of the objective under standard assumptions, and globally\nconverges when the objective function is additionally assumed to satisfy the\nKurdyka- Lojasiewicz (K L) property. We also provide the convergence rate\nwhen the objective satisfies the Lojasiewicz inequality. We apply BIBPA to\nthe symmetric nonnegative matrix tri-factorization (SymTriNMF) problem, where\nwe propose kernel functions for SymTriNMF and provide closed-form solutions for\nsubproblems of BIBPA.\n