A nonlinear method for dimensionality reduction of data using reference nodes

  • Authors:
  • E. V. Myasnikov

  • Affiliations:
  • Image Processing Systems Institute of the Russian Academy of Sciences, Samara, Russia 443001

  • Venue:
  • Pattern Recognition and Image Analysis
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

A nonlinear method for dimensionality reduction based on the hierarchical clusterization of data and the Sammon mapping is proposed in the work. An essential element is the use of lists of reference nodes created based on the results of the hierarchical clusterization of data in the original multidimensional space in the dimensionality reduction. The work quality of the proposed method has been analyzed for a number of feature systems extracted from digital images, as well as for collections of images of various volumes.