What’s the story
A team of scientists from the Yunnan Observatories of the Chinese Academy of Sciences, has developed a neural network-based method for large-scale classification of celestial objects.
The research, published in The Astrophysical Journal Supplement Series, implies the potential of artificial intelligence (AI) in modern astronomy.
The new model can accurately classify stars, galaxies, and quasars by processing both morphological and spectral energy distribution (SED) features simultaneously.
How the model works
Traditional spectroscopic observations, while precise, are time-consuming and resource-intensive.
On the other hand, photometric imaging is more efficient but can lead to ambiguities when classifying objects based solely on morphological or SED features.
For example, the high-redshift quasars and stars both appear as point sources in pictures, making them hard to distinguish.
The new neural network model overcomes these challenges by integrating data from different sources for accurate classification.
Model classified over 27M celestial sources
The neural network model was trained using spectroscopically-confirmed sources from the Sloan Digital Sky Survey Data Release 17.
This training provided a foundation for classification.
When applied to the fifth data release of the Kilo-Degree Survey (KiDS), the model successfully classified more than 27 million celestial sources brighter than r = 23 magnitude across some 1,350 square degrees of sky.
Accuracy tested on Gaia and GAMA data
The model’s performance was validated through extensive testing.
When applied to 3.4 million Gaia sources with a significant proper motion or parallax—traits generally unique to stars—the model correctly identified 99.7% as stellar objects.
A similar success rate was seen with the Galaxy And Mass Assembly Data Release 4, where around 99.7% of sources were accurately classified as either galaxies or quasars.
Model can correct misclassifications in existing catalogs
The research also found that the neural network model could correct the misclassifications in existing catalogs.
Random checks revealed some objects, which were visually identifiable as galaxies but mislabeled as stars in SDSS, were correctly reclassified by the AI.
This highlights the potential of this new method to improve current astronomical databases and enhance our understanding of celestial objects.