作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2012, Vol. 38 ›› Issue (16): 157-160. doi: 10.3969/j.issn.1000-3428.2012.16.040

• 人工智能及识别技术 • 上一篇    下一篇

基于模糊积分的多神经网络集成信息融合

王征宇,肖南峰   

  1. (华南理工大学计算机科学与工程学院,广州 510006)
  • 收稿日期:2011-09-21 修回日期:2012-12-09 出版日期:2012-08-20 发布日期:2012-08-17
  • 作者简介:王征宇(1982-),男,博士研究生,主研方向:神经网络集成,云计算技术;肖南峰,教授、博士
  • 基金资助:
    国家自然科学基金资助项目(61171141)

Multi-neural Network Ensemble Information Fusion Based on Fuzzy Integral

WANG Zheng-yu, XIAO Nan-feng   

  1. (School of Computer Science and Engineering, South China University of Technology, Guangzhou 510006, China)
  • Received:2011-09-21 Revised:2012-12-09 Online:2012-08-20 Published:2012-08-17

摘要: 使用模糊积分实现集成神经网络中的子分类器信息融合,提出一种更加有效和全面的模糊密度,用于模糊积分的计算。以双螺旋分类问题为实验对象,使用集成神经网络实现具有较高正确率的分类方法,对神经网络集成的有效性和各类参数的设定作实验分析,并通过多种模糊密度的比较数据说明该模糊密度函数的有效性。

关键词: 神经网络集成, 集成学习, 信息融合, 模糊积分, 模糊密度, 双螺旋分类问题

Abstract: This paper uses fuzzy integral to complete the information fusion of member classifier in ensemble neural network. A more effective fuzzy density function using in fuzzy integral is proposed. Two spirals classification is regarded as experiment object. Ensemble neural network is used to realize the classification method which has better correct rate. Simulation experiments are done to analyze the validity of neural network and different parameters. And experimental data proves the new fuzzy density function is more effective.

Key words: neural network ensemble, ensemble learning, information fusion, fuzzy integral, fuzzy density, two spirals classification problem

中图分类号: