Beijing – A recent artificial intelligence model, dubbed SpecCLIP, is enabling astronomers to more efficiently analyze stellar data, bridging inconsistencies between datasets collected by different telescopes, Chinese researchers announced Wednesday.
Developed by a team from the National Astronomical Observatories of the Chinese Academy of Sciences (CAS), the University of Chinese Academy of Sciences (UCAS), and other institutions, SpecCLIP acts as a “translator” between spectral data acquired through varying methods, resolutions, and wavelength ranges, according to Huang Yang of UCAS. The model was reported on by the Science and Technology Daily.
Stellar spectra, which contain information about a star’s temperature, chemical composition, and surface gravity, are crucial for tracing the evolutionary history of the Milky Way. However, existing research has been hampered by the difficulty of combining data from projects like China’s Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) and the European Space Agency’s Gaia satellite due to their differing data acquisition techniques.
SpecCLIP addresses this challenge by applying concepts similar to large language models and utilizing a contrastive learning method. This allows the AI to autonomously learn and establish connections between spectral data from diverse sources, effectively converting LAMOST’s low-resolution spectra and Gaia’s high-precision spectra into a “universal language,” Yang explained.
According to a paper published in the Astrophysical Journal, SpecCLIP is not a specialized AI designed for a single task, but rather a foundational model. It is capable of simultaneously predicting stellar atmospheric parameters and elemental abundances, performing spectral similarity searches, and aiding in the identification of unusual celestial objects.
These capabilities are particularly valuable in the field of Galactic archaeology, promising to efficiently sort through massive datasets to identify rare, metal-poor ancient stars. Such stars provide critical evidence for understanding the early formation and merger history of the Milky Way galaxy.
The model has already been applied to ongoing missions, including planet searches. Researchers report that SpecCLIP accurately characterizes the features of planet-hosting stars, improving the efficiency of identifying potentially habitable planets.
