科技创新与应用2025，Vol.15Issue(8)：64-67,4.DOI:10.19981/j.CN23-1581/G3.2025.08.014
基于鲁棒非负矩阵分解的增量式学习研究
摘要
Abstract
Aiming at the phenomenon that the operation scale of robust non-negative matrix factorization continues to increase with the increase of new samples,an incremental learning algorithm for robust non-negative matrix factorization is proposed.The algorithm uses the L2,1 norm to measure the initial and new samples.First,the initial samples are subjected to robust non-negative matrix decomposition,and then the decomposition results are used to participate in subsequent iterative operations.Experiments on ORL and YALE face databases show that compared with the results of the robust non-negative matrix decomposition algorithm and the sparsity restricted non-negative matrix decomposition algorithm,the objective function value of this algorithm reaches balance first when solving,and the best convergence effect is obtained.Save computing time.关键词
增量式学习/鲁棒非负矩阵分解/新增样本/稀疏限制/L2,1范数Key words
incremental learning/robust non-negative matrix factorization/new sample/sparsity constraint/L2,1 norm分类
数理科学引用本文复制引用
杨亮东,赵妍杰,李亚东..基于鲁棒非负矩阵分解的增量式学习研究[J].科技创新与应用,2025,15(8):64-67,4.基金项目
兰州资源环境职业技术大学校级科技项目(X2023A-05) （X2023A-05）
