Automated Classification of ELODIE Stellar Spectral Library Using Probabilistic Artificial Neural Networks

A Probabilistic Neural Network model has been used for automated classification of ELODIE stellar spectral library consisting of about 2000 spectra into 158 known spectro-luminosity classes. The full spectra with 561 flux bins and a PCA reduced set of 57, 26 and 16 components have been used for the training and test sessions. The results shows a spectral type classification accuracy of 3.2 sub-spectral type and luminosity class accuracy of 2.7 for the full spectra and an accuracy of 3.1 and 2.6 respectively with the PCA set. This technique will be useful for future upcoming large data bases and their rapid classification.