Automated morphological classification of apm galaxies by supervised artificial neural networks

We train Artificial Neural Networks to classify galaxies based solely on the morphology of the galaxy images as they appear on blue survey plates. The images are reduced and morphological features such as bulge size and the number of arms are extracted, all in a fully automated manner. The galaxy sample was first classified by 6 independent experts. We use several definitions for the mean type of each galaxy, based on those classifications. We then train and test the network on these features. We find that the rms error of the network classifications, as compared with the mean types of the expert classifications, is 1.8 Revised Hubble Types. This is comparable to the overall rms dispersion between the experts. This result is robust and almost completely independent of the network architecture used.

References(3)

Figures(0)

et al., in preparation) is over 107 CCD images of galaxies. Clearly, such numbers of galaxies cannot be classi ed by humans. There is an obvious need for automated methods that will put the knowledge and experience of the human experts to use and produce very large samples of automatically classi ed galaxies. The rst stage towards achieving this goal was creating a uniform, well-de ned sample to be classi ed by human experts. This was done in previous papers

Naim

, whereby we attempt to teach the ANN to mimic the human classi cations. The ANN is given a set of parameters describing each galaxy and is told what the \correct

Lahav

[5]

In x 6 we give the results of training various con gurations of the ANN, based on di erent choices of ut parameters, mean types and ANN architectures. The