Galaxy Classification by Human Eyes and by Artificial Neural Networks
O. Lahav · arXiv (Cornell University) · 1995
The rapid increase in data on galaxy images at low and high redshift calls for re-examination of the classification schemes and for new automatic objective methods. Here we present a classification method by Artificial Neural Networks. We also show results from a comparative study we carried out using a new sample of 830 APM digitised galaxy images. These galaxy images were classified by 6 experts independently. It is shown that the ANNs can replicate the classification by a human expert almost to the same degree of agreement as that between two human experts, to within 2 $T$-type units. Similar methods can be applied to automatic classification of galaxy spectra. We illustrate it by Principal Component Analysis of galaxy spectra, and discuss future large surveys.