posted on 2023-06-08, 05:22authored byN M Ball, Jonathan LovedayJonathan Loveday, M Fukugita, O Nakamura, S Okamura, J. Brinkmann, R J Brunner
Supervised artificial neural networks are used to predict useful properties of galaxies in the Sloan Digital Sky Survey, in this instance morphological classifications, spectral types and redshifts. By giving the trained networks unseen data, it is found that correlations between predicted and actual properties are around 0.9 with rms errors of order ten per cent. Thus, given a representative training set, these properties may be reliably estimated for galaxies in the survey for which there are no spectra and without human intervention.
Additional authors: Okamura S, Brinkmann J, Brunner R J. This paper demonstrates that supervised artificial neural networks are able to reliably predict Hubble type, spectral type and redshift from standard SDSS galaxy imaging parameters. First author was Loveday's student. Fukugita et al provided training set.