Research on classification of spiral galaxies and stars based on convolutional neural network

Abstract

In astronomy, the shape of galaxies can often reflect the course of its evolution, and can further reveal the development law of the universe. With the progress and development of observation technology, image data has grown explosively, and the efficiency of traditional manual astronomical image classification is far from enough. Image classification based on machine learning effectively solves the problem of low efficiency of artificial classification, but in the face of massive data sets, the classification effect of traditional machine learning is often not as good as deep learning. In this paper, relevant concepts based on neural network image classification algorithm are introduced. Then, according to the characteristics of the mainstream CNN (convolutional neural network) model, the design and optimization of the custom RS CNN model are completed. Finally, the custom RS CNN is used to conduct classification experiments on spiral galaxies and stars.