Unsupervised Machine Learning in Astronomy

This review investigates the application of unsupervised machine learning algorithms to astronomical data. Unsupervised machine learning enables the researchers to analyze large, high-dimensional, and unlabeled data sets and is sometimes considered more helpful for exploratory analysis because it is not limited by present knowledge and can therefore be used to extract new knowledge. Unsupervised machine learning algorithms that have been repeatedly applied to analyze astronomical data are classified according to their usage, including clustering, dimension reduction, and neural network models. This review also discusses anomaly detection and symbolic regression. For each algorithm, this review discusses the algorithm's functioning in mathematical and statistical terms, the algorithm's characteristics (e.g., advantages and shortcomings, possible types of inputs), and the different types of astronomical data analyzed with the algorithm. Example figures are generated. This review aims to provide an up-to-date overview of both the high-level concepts and detailed applications of various unsupervised learning methods in astronomy, highlighting their advantages and disadvantages to help researchers new to unsupervised learning.