Galaxy Image Classification Based on Citizen Science Data: A Comparative Study

Abstract

Many research fields are now faced with huge volumes of data automatically generated by specialised equipment. Astronomy is a discipline that deals with large collections of images difficult to handle by experts alone. As a consequence, astronomers have been relying on the power of the crowds, as a form of citizen science, for the classification of galaxy images by amateur people. However, the new generation of telescopes that will produce images at a higher rate highlights the limitations of this approach, and the use of machine learning methods for automatic classification is considered essential. The goal of this paper is to shed light on the automated classification of galaxy images exploring two distinct machine learning strategies. First, following the classical approach consisting of feature extraction together with a classifier, we compare the state-of-the-art feature extractor for this problem, the WND-CHARM, with our proposal based on autoencoders for feature extraction on galaxy images. We then compare these results with an end-to-end classification using convolutional neural networks. To better leverage the available citizen science data, we also investigate a pre-training scheme that exploits both amateur-and expert-labelled data. Our experiments reveal that autoencoders greatly speed up feature extraction in comparison with WND-CHARM and both classification strategies, either using convolutional neural networks or feature extraction, reach comparable accuracy. The use of pre-training in convolutional neural networks, however, has allowed us to provide even better results.

Assessing Representation Learning and Clustering Algorithms for Computer-Assisted Image Annotation—Simulating and Benchmarking MorphoCluster

“…Despite the dire need for efficient tools to annotate large amounts of image data in many fields [ 1 , 2 , 3 , 4 , 5 , 6 ], a formal exploration of different image annotation approaches is sparse. A formal comparison of different approaches would require additional setup to collect the required metrics and significant effort to annotate the same data multiple times.…”

“…The annotation of images is a central step in many disciplines, including marine ecology [ 1 ], medicine [ 2 , 3 ], astronomy [ 4 ], face recognition [ 5 ] and machine learning [ 6 ]. Considerable progress has been made in the last years regarding the automation of image classification via machine learning approaches, especially since the breakthrough of convolutional neural networks (CNNs) [ 7 ].…”

Transfer Learning and Deep Metric Learning for Automated Galaxy Morphology Representation

“…The majority of labels available in catalogues come from experts inspecting images of galaxies and manually assigning them. The significant increase in the amount of data being made available (in the order of 1e 6 ) has made it increasingly laborious for scientists to classify these galaxies manually citecai2020. One approach is to use machine learning to automate this task [3].…”

“…A consequence of this lack of labelled images and a shortage of experts is citizen-science becoming an increasingly popular approach for assigning labels to images of galaxies [2], [3], [6], [7]. For example, Galaxy Zoo is a widely popular citizen-science, galaxy classification project to reduce the time spent by astronomers manually labelling images of galaxies [3].…”

The Partnership of Citizen Science and Machine Learning: Benefits, Risks, and Future Challenges for Engagement, Data Collection, and Data Quality

“…Astronomy: the involvement of the general public in online astronomy projects started in 2008 with the first release of the Galaxy Zoo project [69]. Traditionally, the classification of galaxy images in Galaxy Zoo was done by citizen scientists, but with advances in ML, the classification task was automated using amateurs and expert labels as input training data [70]. The Milky Way project is another well-known project in this field, with the goal of involving volunteers in identifying bubbles in images collected from space telescopes [71], and to automate the identification, the volunteers' labels were then used to train a random forest algorithm called Brut [72].…”