The team of astronomers used world’s largest deep multi-coloured optical survey, produced by the Pan-STARRS observatory at Haleakalā, Maui, to develop the map.
The team of astronomers used world’s largest deep multi-coloured optical survey, produced by the Pan-STARRS observatory at Haleakalā, Maui, to develop the map.
They fed the algorithm spectroscopic measurements to accurately determine definite object classifications and distances between celestial objects.
The study found that the AI or Machine Learning (ML) approach achieved an overall classification accuracy of 98.1% for galaxies, 97.8% for stars and 96.6% for quasars. And galaxy distance estimates were accurate to almost 3%.
Utilizing a state-of-the-art optimization algorithm, the team leveraged the spectroscopic training set of almost 4 million light sources to teach the neural network to predict source types and galaxy distances, while correcting for light extinction by dust in the Milky Way.
The team said their catalog is double the size of the previous largest map of the universe, created by the Sloan Digital Sky Survey (SDSS), which covers only a third of the sky.
“The new, more accurate, and larger photometric redshift catalog will be the starting point for many future discoveries,” István Szapudi, IfA astronomer and co-author on the study said.
“As Pan-STARRS collects more and more data, we will use machine learning to extract even more information about near-Earth objects, our Solar System, our Galaxy and our Universe,” Ken Chambers, Pan-STARRS Director and IfA Associate Astronomer said.
The resulting 3D catalog is approximately 300 GB in size, and science users can query the catalog through the MAST CasJobs SQL interface, or download the entire collection as a computer-readable table.