Galaxy Detection Machine Learning Project
The correlation between the noise threshold used for detecting a galaxy, and the performance of the binary classifier is studied, to study the process of galaxy detection in telescope images.
Abstract
—Artificial intelligence has proven to be increasingly useful in all types of industries and scientific fields. In particular, machine learning applications can be used for the purpose of object detection and classification in images. In this paper, we use the Square Kilometer Array (SKA) Science Data Challenge 1 (SDC1) data set to train a model that can detect galaxies in images. The importance of this task is in automating the process of galaxy detection in telescope images, especially that the number of galaxies is extremely high in these types of images. For this purpose, we try different approaches based on convolutional neural networks (CNN). The scope of the project covers the classification of images based on the presence or absence of galaxies in them. In this paper, we study the correlation between the noise threshold used for detecting a galaxy, and the performance of our binary classifier.