Using the hypothesis of λ-compactness in the process of forming training sample for forecasting neural network models
Abstract
The problem of forming a qualitative training sample for neural networks in forecasting is solved. The possibility of use the hypothesis of λ-compactness at the step of forming an ensemble of recognizable classes is described. An advanced algorithm of forming a qualitative training sample is offered on the basis of the mechanisms considered.Downloads
Issue
Section
Progressive information technologies, high-efficiency computer systems