Method for stopping of the building process of hierarchical regression in the case of construction of computational procedures of local image processing
Abstract:
The paper deals with the usage of the functional of full cross-validation when solving the problem of automatic building a local signal/images processing procedures, based on empirical data (hierarchical regression). It is proposed a methodology of stopping of the process of formation of various combinations of training and control samples, and the process of constructing of local image processing procedure, based on the interval estimation of the functional of sliding quality control.
Keywords:local processing, hierarchical regression, interval estimation, functional of full cross-validation.
Fulltext:
PDF file (347 kB)