摘要
构造高精度分类模型是对基因表达谱数据分析的主要研究方向之一,但提取不同特征空间产生的分类效果有很大差异,而集成分类系统在一定程度上提高了分类结果的可靠性和稳定性。构建基于PCA和NMF集成分量系统,并基于分析混合矩阵A的hinton图生物学意义建立集成独立分量选择系统,成功运用到基因表达谱分析,实验结果表明,集成分量分类系统优于单个分类器。
One of the main research direction based on gene expression microarray data is construct high-precision classification model.But to extract different features of space-derived classification results are very different.To some extent,ensemble component system improve the relibility and stability of classification results.Construct classification model based on PCA and NMF,and proposed a novel ensemble independent component selection approach based on the biological significance of the hinton graph of the mixing matrix A.This approach applied to gene expression analysis successfully.Experimental tesults demonstrate that ensembles component system is better than that of single classification system.
出处
《电子设计工程》
2012年第7期9-12,共4页
Electronic Design Engineering
基金
国家自然科学基金项目(60801060)
上海市教委科研创新项目(11YZ141)
上海市科委青年科技启明星计划(A类)(11QA1402900)
关键词
微阵列数据
NMF
ICA
集成分类
microarray data
Non-gegative Matrix Factorization（NMF）
Independent Component Analysis（ICA）
ensemble classification
