Kernel classification with missing data and the choice of smoothing parameters

被引：0
作者：
Levon Demirdjian
Majid Mojirsheibani
机构：
[1] University of California,Department of Statistics
[2] California State University,Department of Mathematics
来源：
Statistical Papers | 2019年 / 60卷
关键词：
Classification; Kernel; Missing covariate; Consistency; Shatter coefficient;
D O I：
暂无
中图分类号：
学科分类号：
摘要：
Methods are proposed for selecting smoothing parameters of kernel classifiers in the presence of missing covariates. Here the missing covariates can appear in both the data and in the unclassified observation that has to be classified. The proposed methods are quite straightforward to implement. Exponential performance bounds will be derived for the resulting classifiers. Such bounds, in conjunction with the Borel–Cantelli lemma, provide various strong consistency results. Several numerical examples are presented to illustrate the effectiveness of the proposed procedures.
引用
收藏
页码：1487 / 1513
页数：26
相关论文
共 35 条
  • [1] Chung H(2000)Discriminant analysis when a block of observations is missing Ann Inst Stat Math 52 544-556
  • [2] Han C(1981)On the almost everywhere convergence of nonparametric regression function estimates Ann Stat 9 1310-1319
  • [3] Devroye L(1982)Bounds for the uniform deviation of empirical measures J Multivar Anal 12 72-79
  • [4] Devroye L(1989)An equivalence theorem for J Stat Plann Inference 23 71-82
  • [5] Devroye L(1980) convergence of the kernel regression estimate Ann Stat 8 231-239
  • [6] Krzyzak A(1982)Distribution-free consistency results in nonparametric discrimination and regression function estimation Handb Stat 2 193-197
  • [7] Devroye L(1973)Nearest neighbor methods in discrimination Biometrics 29 241-256
  • [8] Wagner T(1978)Sample-based multinomial classification Ann Stat 6 515-533
  • [9] Devroye L(1980)Asymptotically efficient solutions to the classification problem J Multivar Anal 10 611-627
  • [10] Wagner T(2005)Consistent nonparametric regression from recursive partitioning schemes Ann Stat 33 284-306
1234