An overview of ML in astrophysics
Viviana Acquaviva, New York City College of Technology, CUNY
Astronomy has truly become a "Big Data" science in the last few decades, thanks to technological advances in telescope development and the availability of space-based observations. The dramatic improvement in the volume and quality of available data gives us a chance to address crucial questions, but requires to us to adequately expand the set of tools used for data analysis, interpretation, and storage.
Machine learning techniques are found to be increasingly useful in solving a variety of problems, from automatic classification of galaxy morphology, to photometric redshifts, to source identification in crowded fields, to estimation of star and galaxy properties, to outlier recognition and anomaly detection, to dimensionality reduction, just to quote a few examples.
In this talk, I will review some of the most notable applications of machine learning and deep learning to the analysis of astronomical data, and describe how mastering these techniques can help astronomers extract maximal information from high-volume, rich data, such as those from the upcoming Vera Rubin Observatory survey. If time allows it, I will present a few specific examples drawn from my research experience analyzing galaxy spectra and training models on cosmological simulations.