ON EXPONENTIAL BOUNDS ON THE BAYES RISK OF THE KERNEL CLASSIFICATION RULE

被引：8
|
作者：
KRZYZAK, A
机构：
[1] Department of Computer Science, Concordia University, Montréal, H3G, 1M8
来源：
IEEE TRANSACTIONS ON INFORMATION THEORY | 1991年 / 37卷 / 03期
基金：
加拿大自然科学与工程研究理事会;
关键词：
KERNEL CLASSIFICATION RULE; L1 STRONG CONSISTENCY; RATES OF STRONG CONVERGENCE; EXPONENTIAL INEQUALITY;
D O I：
10.1109/18.79905
中图分类号：
TP [自动化技术、计算机技术];
学科分类号：
0812 ;
摘要：
Improved exponential bounds on Bayes probability of error for the nonparametric kernel classification rule are derived. It is shown, using the martingale device, that weak, strong, and complete L1 Bayes risk consistencies are equivalent. Consequently the conditions on the smoothing sequence h(n) --> 0 and nh(n) --> infinity are necessary and sufficient for Bayes rish consistency of the kernel classification rule. The rate of convergence of the kernel classification rule is also given.
引用
收藏
页码：490 / 499
页数：10
相关论文
共 50 条
12345