Complexity of Functional Learning on Some Classes of Multivariate Functions

被引：0
|
作者：
Ye, Peixin [1 ]
He, Qing [2 ]
机构：
[1] Nankai Univ, Sch Math Sci, Tianjin 300071, Peoples R China
[2] Informat Proc ICT CAS, Key Lan Intelligent, Beijing, Peoples R China
来源：
ICNC 2008: FOURTH INTERNATIONAL CONFERENCE ON NATURAL COMPUTATION, VOL 3, PROCEEDINGS | 2008年
关键词：
functional learning; computational complexity; randomized methods; anisotropic classes;
D O I：
10.1109/ICNC.2008.225
中图分类号：
TP18 [人工智能理论];
学科分类号：
081104 ; 0812 ; 0835 ; 1405 ;
摘要：
We study the error of the functional learning on anisotropic Sobolev classes W-p(r)(I-d) and Holder-Nikolskii classes H-p(r)(I-d) with respect to the worst case randomized methods and the average case deterministic methods, where 1 <= P <= infinity. Our results show that if p >= 2 then the stochastic and average error bounds are essentially smaller than the deterministic ones. Quantitatively the improvement amounts to the factor n(-1/2).
引用
收藏
页码：141 / +
页数：2
相关论文
共 50 条
12345