http://arxiv.org/abs/2209.12194
This paper follows series of our works on the applicability of various machine learning methods to the morphological galaxy classification (Vavilova et al., 2021, 2022). We exploited the sample of 315776 SDSS DR9 galaxies with absolute stellar magnitudes of -24m<Mr<-19.4m at 0.003<z<0.1 as a target data set for the CNN classifier based on the DenseNet-201. Because it is tightly overlapped with the Galaxy Zoo 2 (GZ2) sample, we use these annotated data as the training data set to classify galaxies into 34 detailed features. In the presence of a pronounced difference of visual parameters between galaxies from the GZ2 training data set and galaxies without known morphological parameters, we applied novel procedures, which allowed us for the first time to get rid of this difference for smaller and fainter SDSS galaxies.
We describe in detail the adversarial validation technique as well as how we managed the optimal train-test split of galaxies from the training data set. We have also found optimal galaxy image transformations to increase the classifier generalization ability. It can be considered as another way to improve the human bias for those galaxy images that had a poor vote classification in the GZ project. Such an approach, likely auto-immunization, when the CNN classifier trained on very good images is able to retrain bad images from the same homogeneous sample, can be considered co-planar to other methods of combating the human bias.
The accuracy of CNN classifier is in the range of 83.3-99.4 percent depending on 32 features. As a result, for the first time, we assigned the detailed morphological classification for more than 140K low-redshift galaxies, especially at the fainter end. We accentuate on the typical problem points of galaxy CNN image classification from the astronomical point of view. The catalogs will be available through the VizieR.
V. Khramtsov, I. Vavilova, D. Dobrycheva, et. al.
Tue, 27 Sep 22
9/89
Comments: 42 pages, 4 tables, 13 figures, 122 references