Classifying Seyfert Galaxies with Deep Learning

Traditional classification for subclass of the Seyfert galaxies is visual inspection or using a quantity defined as flux ratio between Balmer line and forbidden line. One algorithm deep learning Convolution Neural Network (CNN) has shown successful results. We building 1-dimension CNN model to distinguish 1.9 spectra from 2 galaxies. find our can recognize with an accuracy over 80% pick out additional sample which was missed by inspection. use new improve performance obtain 91% precision 1.9. These results indicate among spectra. decompose H{\alpha} emission fitting Gaussian components derive width flux. velocity distribution broad component extending tail toward higher end luminosity slightly weaker than original sample. This result indicates that sources have relatively weak component. Besides, we check distributions host galaxy morphology samples dominant large bulge galaxy. In end, present online catalog 1297 measurement