Learning from uniformly ergodic Markov chains

被引:30
作者
Zou, Bin [1 ,2 ]
Zhang, Hai [2 ]
Xu, Zongben [2 ]
机构
[1] Hubei Univ, Fac Math & Comp Sci, Wuhan 430062, Peoples R China
[2] Xi An Jiao Tong Univ, Fac Sci, Inst Informat & Syst Sci, Inst Informat & Syst Sci, Xian 710049, Peoples R China
关键词
ERM algorithms; Uniform ergodic Markov chain samples; Generalization bound; Uniform convergence; Relative uniform convergence; DEPENDENT OBSERVATIONS; MIXING SEQUENCES;
D O I
10.1016/j.jco.2009.01.001
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Evaluation for generalization performance of learning algorithms has been the main thread of machine learning theoretical research. The previous bounds describing the generalization performance of the empirical risk minimization (ERM) algorithm are usually established based on independent and identically distributed (i.i.d.) samples. In this paper we go far beyond this classical framework by establishing the generalization bounds of the ERM algorithm with uniformly ergodic Markov chain (u.e.M.c.) samples. We prove the bounds on the rate of uniform convergence/relative uniform convergence of the ERM algorithm with u.e.M.c. samples, and show that the ERM algorithm with u.e.M.c. samples is consistent. The established theory underlies application of ERM type of learning algorithms. (C) 2009 Elsevier Inc. All rights reserved.
引用
收藏
页码:188 / 200
页数:13
相关论文
共 24 条
[11]   Hoeffding's inequality for uniformly ergodic Markov chains [J].
Glynn, PW ;
Ormoneit, D .
STATISTICS & PROBABILITY LETTERS, 2002, 56 (02) :143-146
[12]  
HASTINGS WK, 1970, BIOMETRIKA, V57, P97, DOI 10.1093/biomet/57.1.97
[13]  
JONES G. L., 2004, Probability Surveys, V1, P299, DOI 10.1214/154957804100000051
[14]  
Meyn S.P., 1993, MARKOV CHAINS STOCHA
[15]   Minimum complexity regression estimation with weakly dependent observations [J].
Modha, DS ;
Masry, E .
IEEE TRANSACTIONS ON INFORMATION THEORY, 1996, 42 (06) :2133-2145
[16]   Shannon sampling and function reconstruction from point values [J].
Smale, S ;
Zhou, DX .
BULLETIN OF THE AMERICAN MATHEMATICAL SOCIETY, 2004, 41 (03) :279-305
[17]  
SMALE S, ONLINE LEARNING MARK
[18]   Learning from dependent observations [J].
Steinwart, Ingo ;
Hush, Don ;
Scovel, Clint .
JOURNAL OF MULTIVARIATE ANALYSIS, 2009, 100 (01) :175-194
[19]  
Vapnik V.N., 1998, STAT LEARNING THEORY, V1
[20]  
Vidyasagar Mathukumalli, 2002, Learning and Generalization with Applications to Neural Networks