Application of Artificial Intelligence in Astronomy and Science

by Marwan Gebrane, associate professor of Chemistry and Physics

Innovation in technological capabilities and processes, as well as an associated increase of astronomical data, presents the need for a more efficient way to process this data. Deep Learning (DL) and Machine (ML) Learning have progressed to accommodate these needs and have proven to be useful techniques to build effective models for the analysis of this new data. In this specific case, Dr. Gebran and his undergraduate researchers are utilizing Neural Networks within ML/DL to create a better system for use in stellar spectroscopy. They are building a network to classify and characterize stars by temperature, rotation, gravity, content of metal and other astrophysical properties.  Characterizing and classifying the stars is the first step in any stellar physics project (e.g. Chemical evolution of the galaxy, peculiar stars detection, exoplanet’s host star properties...)

These kinds of networks can be tweaked and applied to any complex classification/regression problem in science. For example, in collaboration with Dr. Bentley these same techniques are being applied to count and track bats flying in videos.