基于二阶段迭代的非负矩阵分解的分类模型  

2-STGNMF:Supervised 2-Stage Iterative Nonnegative Matrix Factorization Model for General Classification

作　　者：全聪 李晨亮[2] 吴黎兵[1,3] QUAN Cong;LI Chenliang;WU Libing(School of Computer Science,Wuhan University,Wuhan 430072,Hubei,China;School of Cyber Science and Engineering,Wuhan University,Wuhan 430070,Hubei,China;Shenzhen Research Institute,Wuhan University,Shenzhen 518057,Guangdong,China)

机构地区：[1]武汉大学计算机学院,湖北武汉430072 [2]武汉大学国家网络安全学院,湖北武汉430070 [3]武汉大学深圳研究院,广东深圳518057

出　　处：《武汉大学学报（理学版）》2020年第2期190-196,共7页Journal of Wuhan University:Natural Science Edition

基　　金：国家自然科学基金(61772377,61572370,91746206);湖北省自然科学基金(2017CFA007);深圳市科技计划项目(JCYJ20170818112550194)。

摘　　要：为了有效地结合标签信息与非负矩阵分解技术,提升现有的非负矩阵分解算法划分数据的性能,提出一种用于分类问题的基于二阶段迭代的非负矩阵分解模型(2-stage iterative nonnegative matrix factorization model,2-STGINMF),在阶段1,基于训练样本之间的关系和标签信息,用非负矩阵分解技术学习训练样本的置信度分布矩阵。在阶段2,根据训练样本的置信度分布矩阵,基于训练样本和测试样本之间的关系以及测试样本内部的关系,学习测试样本关于不同类别的置信度分布矩阵。此外,提出了一种迭代式训练机制解决标签稀疏性的问题。实验结果表明,与当前的一些机器学习方法和矩阵分解方法相比,本文提出的2-STGINMF模型在不同类型的数据分类问题上都达到了最优的性能且在训练样本较少时也能获得较好的分类结果。In this paper,we propose a supervised 2-stage iterative nonnegative matrix factorization model for general classification(2-STGINMF),which aims to effectively incorporate the label information into non-negative matrix factorization to further enhance the performance of nonnegative matrix factorization on data division.At stage one,based on the relations within training samples and the label information,we learn the confidence distribution matrix by non-negative matrix factorization.At stage two,based on the relations between test and training samples and the relationship within the test sample,we learn the confidence distribution of test samples towards different classes.In addition,we propose an iterative training mode to alleviate the scarcity of labeled data.The experimental results show that compared with existing machine learning methods and matrix decomposition methods,the proposed 2-STGINMF model achieves the optimal performance in different datasets,and superior classification performance especially when the training data is small.

关 键 词：监督学习 分类 非负矩阵分解 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]