首页> 外文期刊>Information Theory, IEEE Transactions on >Conditions for Existence of Uniformly Consistent Classifiers
【24h】

Conditions for Existence of Uniformly Consistent Classifiers

机译:一致一致分类器的存在条件

获取原文
获取原文并翻译 | 示例
           

摘要

We consider the statistical problem of binary classification, which means attaching a random observation X from a separable metric space E to one of the two classes, 0 or 1. We prove that the consistent estimation of conditional probability p(X)= mathsf {P}(Y=1 mid X) , where Y is the true class of X , is equivalent to the consistency of a class of empirical classifiers. We then investigate for what classes P there exist an estimate hat p that is consistent uniformly in pin P . We show that this holds if and only if P is a totally bounded subset of L^{1}(E,mu ) , where mu is the distribution of X . In the case, where E is countable, we give a complete characterization of classes Pi , allowing consistent estimation of p , uniform in (mu ,p)in Pi .
机译:我们考虑二进制分类的统计问题,这意味着将来自可分离度量空间E的随机观测值X附加到两个类别0或1中。我们证明了条件概率p(X)= mathsf {P的一致估计}(Y = 1 X中间),其中Y是X的真实类,它等效于一类经验分类器的一致性。然后,我们调查对于哪种类别P,存在一个在引脚P中一致的估计帽子p。我们证明,当且仅当P是L ^ {1}(E,mu)的一个完全有界子集时,这成立,其中mu是X的分布。在E是可数的情况下,我们给出了Pi类的完整表征,允许对p进行一致的估计,在Pi中的(mu,p)是均匀的。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号