How tracer objects can improve competitive learning algorithms in astronomy

M. Hernandez-Pajares, J. Floris, F. Murtagh

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

The main objective of this paper is to discuss how the use of tracer objects in competitive learning can improve results in stellar classification. To do this, we work with a Kohonen network applied to a reduced sample of the Hipparcos Input Catalogue, which contains missing values. The use of synthetic stars as tracer objects allows us to determine the discrimination quality and to find the best final values of the cluster centroids, or neuron weights.

Original languageEnglish
Pages (from-to)317-330
Number of pages14
JournalVistas in Astronomy
Volume38
Issue numberPart 3
DOIs
Publication statusPublished - 1994
Externally publishedYes

Fingerprint Dive into the research topics of 'How tracer objects can improve competitive learning algorithms in astronomy'. Together they form a unique fingerprint.

  • Cite this