Linear and nonlinear mapping of patterns

作者:

Highlights:

摘要

Discrete Karhunen-Loeve-transform and a newly developed nonlinear mapping are used to obtain a structure-preserving mapping of high-dimensional data to a plane. It seems that the potential power of nonlinear mappings is needed only for artificial data.

论文关键词:Linear mapping,Nonlinear mapping,Cluster analysis,handprinted numerals,Speaker identification,Word recognition

论文评审过程:Received 26 February 1979, Revised 3 July 1979, Available online 19 May 2003.

论文官网地址:https://doi.org/10.1016/0031-3203(80)90006-0