Learning to recognize patterns with a probabilistic teacher
作者：
Highlights：
•
摘要
A problem of learning with a probabilistic teacher is considered. Neither prior class probabilities nor class densities are assumed to be known and pattern recognition procedures are derived from nonparametric density and regression estimates. Weak and strong Bayes risk consistency of the procedures is shown. Examples of procedures using the kernel, the nearest neighbor and the orthogonal series estimates are given.
论文关键词：Pattern recognition,Classification,Learning,Probabilistic teacher,Imperfect teacher,Density estimation,Regression estimation,Nonparametric estimation
论文评审过程：Received 1 February 1979, Revised 2 October 1979, Accepted 10 December 1979, Available online 19 May 2003.
论文官网地址：https://doi.org/10.1016/0031-3203(80)90040-0
