Greedy algorithms and M-term approximation with regard to redundant dictionaries

被引:48
作者
Temlyakov, VN [1 ]
机构
[1] Univ S Carolina, Dept Math, Columbia, SC 29208 USA
基金
美国国家科学基金会;
关键词
D O I
10.1006/jath.1998.3265
中图分类号
O1 [数学];
学科分类号
0701 ; 070101 ;
摘要
We study the efficiency of greedy type algorithms with regard to redundant dictionaries in Hilbert space and we prove a general result which gives a sufficient condition on a dictionary to guarantee that the pure greedy algorithm is near best in the sense of power decay of error of approximation. We discuss also some important examples. It is already known (see DeVore and Temlyakov, Adv. Comput. Math. 5 (1996), 173-187) that the Pure Greedy Algorithm for some dictionaries has a saturation property. We construct an example which shows that a natural generalization of the Pure Greedy Algorithm also has a saturation property. Next we discuss some new phenomena which occur in approximation by a greedy type algorithm with regards to a highly redundant dictionary. (C) 1999 Academic Press.
引用
收藏
页码:117 / 145
页数:29
相关论文
共 11 条
[1]   UNIVERSAL APPROXIMATION BOUNDS FOR SUPERPOSITIONS OF A SIGMOIDAL FUNCTION [J].
BARRON, AR .
IEEE TRANSACTIONS ON INFORMATION THEORY, 1993, 39 (03) :930-945
[2]  
DARKEN C, 1993, 6TH P ANN ACM C COMP, P303
[3]   Adaptive greedy approximations [J].
Davis G. ;
Mallat S. ;
Avellaneda M. .
Constructive Approximation, 1997, 13 (1) :57-98
[4]   COMPRESSION OF WAVELET DECOMPOSITIONS [J].
DEVORE, RA ;
JAWERTH, B ;
POPOV, V .
AMERICAN JOURNAL OF MATHEMATICS, 1992, 114 (04) :737-785
[5]   Nonlinear approximation in finite-dimensional spaces [J].
DeVore, RA ;
Temlyakov, VN .
JOURNAL OF COMPLEXITY, 1997, 13 (04) :489-508
[6]   Some remarks on greedy algorithms [J].
DeVore, RA ;
Temlyakov, VN .
ADVANCES IN COMPUTATIONAL MATHEMATICS, 1996, 5 (2-3) :173-187
[7]  
Donahue MJ, 1997, CONSTR APPROX, V13, P187
[8]  
Dubinin V. V., 1997, THESIS U S CAROLINA
[9]   A SIMPLE LEMMA ON GREEDY APPROXIMATION IN HILBERT-SPACE AND CONVERGENCE-RATES FOR PROJECTION PURSUIT REGRESSION AND NEURAL NETWORK TRAINING [J].
JONES, LK .
ANNALS OF STATISTICS, 1992, 20 (01) :608-613
[10]   The best m-term approximation and greedy algorithms [J].
Temlyakov, VN .
ADVANCES IN COMPUTATIONAL MATHEMATICS, 1998, 8 (03) :249-265