Sparse Deep Nonnegative Matrix Factorization

作     者：Zhenxing Guo Shihua Zhang Zhenxing Guo;Shihua Zhang

作者机构：the Academy of Mathematics and Systems ScienceChinese Academy of Sciences(CAS)Beijing 100190China NCMISCEMSRCSDSAcademy of Mathematics and Systems ScienceChinese Academy of SciencesBeijing 100190China the School of Mathematical SciencesUniversity of Chinese Academy of SciencesBeijing 100049China 

出 版 物：《Big Data Mining and Analytics》 (大数据挖掘与分析（英文）)

年 卷 期：2020年第3卷第1期

页      面：13-28页

核心收录：

学科分类：0810[工学-信息与通信工程] 1205[管理学-图书情报与档案管理] 07[理学] 0839[工学-网络空间安全] 070104[理学-应用数学] 0811[工学-控制科学与工程] 0701[理学-数学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：supported by the National Natural Science Foundation of China(Nos.11661141019 and 61621003) the National Ten Thousand Talent Program for Young Topnotch Talents Chinese Academy Science(CAS)Frontier Science Research Key Project for Top Young Scientist(No.QYZDB-SSW-SYS008) the Key Laboratory of Random Complex Structures and Data Science,CAS(No.2008DP173182) 

主　　题：sparse Nonnegative Matrix Factorization(NMF) deep learning Nesterov’s accelerated gradient algorithm 

摘      要：Nonnegative Matrix Factorization(NMF)is a powerful technique to perform dimension reduction and pattern recognition through single-layer data representation ***,deep learning networks,with their carefully designed hierarchical structure,can combine hidden features to form more representative features for pattern *** this paper,we proposed sparse deep NMF models to analyze complex data for more accurate classification and better feature *** models are designed to learn localized features or generate more discriminative representations for samples in distinct classes by imposing L1-norm penalty on the columns of certain *** extending a one-layer model into a multilayer model with sparsity,we provided a hierarchical way to analyze big data and intuitively extract hidden features due to *** adopted the Nesterov’s accelerated gradient algorithm to accelerate the computing *** also analyzed the computing complexity of our frameworks to demonstrate their *** improve the performance of dealing with linearly inseparable data,we also considered to incorporate popular nonlinear functions into these frameworks and explored their *** applied our models using two benchmarking image datasets,and the results showed that our models can achieve competitive or better classification performance and produce intuitive interpretations compared with the typical NMF and competing multilayer models.