基于模糊随机样本的结构风险最小化原则

计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (34): 51-55.

基于模糊随机样本的结构风险最小化原则

何其慧1，姚登宝2，王翠翠2，毛军军2，3

1.安徽经济管理学院，合肥 230059
2.安徽大学数学科学学院，合肥 230039
3.安徽大学计算智能与信号处理教育部重点实验室，合肥 230039

收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-12-01 发布日期:2011-12-01

Structural risk minimization principle based on fuzzy-stochastic samples

HE Qihui1，YAO Dengbao2，WANG Cuicui2，MAO Junjun2，3

1.Anhui Institute of Economic Management，Hefei 230059，China
2.College of Mathematical Sciences，Anhui University，Hefei 230039，China
3.Key Lab of Intelligent Computing & Signal Processing of MOE，Institute of Artificial Intelligence，Anhui University，Hefei 230039，China

Received:1900-01-01 Revised:1900-01-01 Online:2011-12-01 Published:2011-12-01

摘要/Abstract

摘要： 基于模糊随机样本，提出了熵、退火熵、生长函数和VC维等概念，并在此基础上构建了基于VC维的学习过程一致收敛速度的界;给出了基于模糊随机样本的结构风险最小化原则（FSSRM原则），证明了基于FSSRM原则下收敛速度渐进分析的相关性质。

关键词: 模糊随机样本, VC维, 结构风险最小化原则, 渐进分析

Abstract: The concepts of entropy，annealed entropy，growth function and VC dimension are proposed based on fuzzy-
stochastic samples，and on the basis of these，the bounds on the rate of uniform convergence of learning process based on VC dimension are constructed.At same time，the idea of the structural risk minimization principle based fuzzy-stochastic samples has been presented，relative properties of asymptotic analysis about convergent rate based on FSSRM principle have been also proved.

Key words: fuzzy-stochastic samples, VC dimension, the structural risk minimization principle, asymptotic analysis.

何其慧1，姚登宝2，王翠翠2，毛军军2，3. 基于模糊随机样本的结构风险最小化原则[J]. 计算机工程与应用, 2011, 47(34): 51-55.

HE Qihui1，YAO Dengbao2，WANG Cuicui2，MAO Junjun2，3. Structural risk minimization principle based on fuzzy-stochastic samples[J]. Computer Engineering and Applications, 2011, 47(34): 51-55.

[1]	姚程宽1，许建华2. 双正则化参数的L2-SVM参数选择[J]. 计算机工程与应用, 2014, 50(8): 99-102.
[2]	张植明¹，田景峰². 基于随机粗糙样本的结构风险最小化原则[J]. 计算机工程与应用, 2010, 46(21): 51-54.
[3]	白云超¹，张植明². 拟概率空间上结构风险最小化原则[J]. 计算机工程与应用, 2010, 46(20): 37-39.
[4]	何渊淘，邓伟. 改进的不均衡样本集支持向量机预处理方法[J]. 计算机工程与应用, 2010, 46(10): 36-37.
[5]	张植明. 双重随机样本的结构风险最小化原则[J]. 计算机工程与应用, 2009, 45(1): 51-55.