电子学报 ›› 2016, Vol. 44 ›› Issue (10): 2308-2313.DOI: 10.3969/j.issn.0372-2112.2016.10.004

• 学术论文 • 上一篇    下一篇

用于微阵列数据分类的子空间融合演化超网络

王进, 刘彬, 张军, 陈乔松, 邓欣   

  1. 重庆邮电大学计算智能重庆市重点实验室, 重庆 400065
  • 收稿日期:2015-03-11 修回日期:2015-06-30 出版日期:2016-10-25
    • 通讯作者:
    • 刘彬
    • 作者简介:
    • 王进,男,1979年1月出生于重庆,教授.主要研究方向为数据挖掘、机器学习.E-mail:wangjin@cqupt.edu.cn
    • 基金资助:
    • 国家自然科学基金 (No.61203308,No.61403054); 重庆教委科学技术研究项目 (自然科学类) (No.KJ1400436); 重庆市基础与前沿研究计划项目 (No.cstc2014jcyjA40001)

Classification of Microarray Data Using Evolutionary Hypernetworks with Subspace Fusion

WANG Jin, LIU Bin, ZHANG Jun, CHEN Qiao-song, DENG Xin   

  1. Chongqing Key Laboratory of Computational Intelligence, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
  • Received:2015-03-11 Revised:2015-06-30 Online:2016-10-25 Published:2016-10-25

摘要:

针对传统模式识别方法在学习具有小样本特性的DNA微阵列数据时存在的过拟合问题,本文提出了一种子空间融合演化超网络模型.该模型通过子空间划分、超边全覆盖和子空间融合三种方法降低模型对初始化的依赖,减少了对数据空间的拟合误差,提高了演化超网络的泛化能力.对四个DNA微阵列数据集的实验结果表明,子空间融合演化超网络的识别率和在小样本训练集下的泛化能力均优于参与对比的其他传统模式识别方法.

关键词: 模式识别, 微阵列数据分类, 演化超网络, 子空间, 过拟合

Abstract:

In order to solve the over-fitting problem of the traditional pattern recognition approaches under the DNA microarray data with small train samples,a subspace fusion-based evolutionary hypernetwork model is proposed in this paper.With the methods of subspace division,hyperedge coverage,and subspace fusion,the proposed scheme reduces the dependence on the initialization,decreases the fitting error of the data space,and enhances the generalization ability of the evolutionary hypernetwork.The experimental results on four DNA microarray datasets show that the proposed model achieves higher classification accuracy and stronger generalization ability than other compared traditional pattern recognition method.

Key words: pattern recognition, microarray data classification, evolutionary hypernetwork, subspace, over-fitting

中图分类号: