Computer Engineering and Applications ›› 2010, Vol. 46 ›› Issue (16): 245-248.DOI: 10.3778/j.issn.1002-8331.2010.16.071

• 工程与应用 • Previous Articles    

NMF-based method for data classification

ZHANG Zhong-yuan1,ZHANG Xiang-sun2   

  1. 1.School of Statistics,Central University of Finance and Economics,Beijing 100081,China
    2.Academy of Mathematics and Systems Science,Chinese Academy of Sciences,Beijing 100190,China
  • Received:2008-12-09 Revised:2009-05-04 Online:2010-06-01 Published:2010-06-01
  • Contact: ZHANG Zhong-yuan



  1. 1.中央财经大学 统计学院,北京 100081
    2.中国科学院 数学与系统科学研究院,北京 100190
  • 通讯作者: 张忠元

Abstract: In bioinformatics,an important task is to classify tumor samples into different classes based on microarray technology which enables people to monitor entire genome in a single chip using a system’s approach.The key difficulty of this problem,compared with many traditional classification problems,is the high dimensionality in gene space and the small number of samples that will be classified.Non-negative Matrix Factorization(NMF) has coped with this difficulty successfully in microarray data clustering.NMF is extended to tumor classification and the result shows its competition.NMF-based method has three advantages:Good classification performance,parameter-independent and good interpretability.

Key words: Non-negative Matrix Factorization(NMF), microarray, data classification

摘要: 在生物信息学中,一个重要的问题是基于微芯片技术将肿瘤分类到不同的类别中去。和许多传统的分类问题相比,这个问题的主要困难是基因空间的维数很高,而要分类的样本数量很小。非负矩阵分解(NMF)在微芯片数据聚类问题中已经成功地解决了这个问题。将非负矩阵分解拓展到数据分类,尤其是肿瘤分类中去取得了很好的效果。基于非负矩阵分解的方法有三个优点:良好的分类成绩,无参数和良好的可解释性。

关键词: 非负矩阵分解, 微芯片, 数据分类

CLC Number: