低秩张量填充的加速随机临近梯度算法  

AN ACCELERATED STOCHASTIC PROXIMAL GRADIENT ALGORITHM FORLOW RANK TENSOR COMPLETION

作　　者：郭雄伟 王川龙[1] Guo Xiongwei;Wang Chuanlong(Key Laboratory of Engineering and Computational Science(Taiyuan Normal University),Shanci Province Department of Education,Jinzhong 030619,China)

机构地区：[1]工程科学计算山西省高等学校重点实验室(太原师范学院),晋中030619

出　　处：《计算数学》2022年第4期534-544,共11页Mathematica Numerica Sinica

基　　金：国家自然科学基金(11371275);山西省自然科学基金(201601D011004);山西省研究生教育创新项目(2021Y713);太原师范学院研究生教育创新项目(SYYJSJC-2164)资助。

摘　　要：本文提出了一种求解低秩张量填充问题的加速随机临近梯度算法张量填充模型可以松弛为平均组合形式的无约束优化问题,在迭代过程中,随机选取该组合中的某一函数进行变量更新,有效减少了张量展开、矩阵折叠及奇异值分解带来的较大的计算花费。本文证明了算法的收敛率为O(1/k^(2)).最后,随机生成的和真实的张量填充实验结果表明新算法在CPU时间上优于现有的三种算法.In this paper,an accelerated stochastic proximal gradient algorithm is proposed for low rank tensor completion problem.The tensor completion model is relaxed into an unconstrained optimization problem in the form of average combination.In the iterative process,a function of this combination is randomly selected for variable update,which effectively reduces the large calculation cost caused by tensor unfolding,matrix folding,and singular value decomposition.We prove that the convergence rate of this algorithm is O(1/k^(2)).Finally,numerical results on randomly generated and real tensor completion problems demonstrate new algorithm is better than three existing algorithms in CPU time.

关 键 词：张量填充 加速临近梯度算法. 

分 类 号：O183.2[理学—数学]