Classifying stars: A comparison between classical, genetic and neural networks algorithms..
M. Hernández-Pajares, F.Comellas , E. Monte*, J. Floris
Departament de Matemàtica Aplicada i Telemàtica;
Universitat Politècnica de Catalunya
*Departament de Teoria del Senyal i Comunicacions;
Universitat Politècnica de Catalunya
pp. 325--330 of Astronomy from Large Databases II;
Edited by: A. Heck and F. Murtagh.
European Southern Observatory, 1992.
ISBN 3-923524-47-1
One of the relevant studies that are carried out
with stellar samples is
the segregation of stars in populations with the aid of
spectral,
photometric and/or kinematic
data. We present the first results of the use of four different
classification techniques on stellar catalogues: the Self-Organizing Map and
Multi-Layer Perceptron, two different neural network architectures, and the
Genetic
algorithms and Hierarchical Clustering. Also the Principal Component
Analysis is applied to the initial data that consist of two
synthetic samples of the
Solar Neighborhood with 3-D position and velocity, metallicity and age.
Load: