A Clearer View of the Milky Way with Gaia and Machine Learning

A group of scientists led by the Leibniz Institute for Astrophysics Potsdam (AIP) and the Institute of Cosmos Sciences of the University of Barcelona (ICCUB) used a new machine learning model to process data from 217 millions of stars observed by the Gaia mission in an extremely efficient manner.

The results are competitive with traditional methods used to estimate stellar parameters. This new approach opens exciting opportunities to map features such as interstellar extinction and metallicity across the Milky Way, contributing to the understanding of stellar populations and the structure of our galaxy.

With the third data release from the European Space Agency’s Gaia space mission, astronomers have gained access to improved measurements for 1.8 billion stars, providing a large amount of data for studying the Milky Way.

However, effectively analyzing such a large dataset presents challenges. In the study, researchers explored the use of machine learning to estimate key stellar properties using spectrophotometric data from Gaia. The model was trained on high-quality data from 8 million stars and achieved reliable predictions with low uncertainties.

The work is published in the journal Astronomy and astrophysics.

“The underlying technique, called extreme gradient trees, allows precise stellar properties, such as temperature, chemical composition, and obscuration of interstellar dust, to be estimated with unprecedented efficiency. The learning model automatically developed, SHBoost, accomplishes its tasks, including model training and prediction, in four hours on a single GPU, a process that previously required two weeks and 3,000 high-performance processors,” says Arman Khalatyan of AIP and first author of the study.

“The machine learning method thus significantly reduces calculation time, energy consumption and CO emissions.₂ broadcast.” This is the first time such a technique has been successfully applied to stars of all types at once.

The model trains on high-quality spectroscopic data from smaller stellar surveys, then applies this learning to Gaia’s third large data release (DR3), extracting key stellar parameters using only photometric and astrometric data , as well as Gaia’s low-resolution XP spectra.

“The high quality of the results reduces the need for additional resource-intensive spectroscopic observations when searching for good candidates to select for further studies, such as rare metal-poor or supermetal-rich stars, crucial for understanding early phases . of the formation of the Milky Way”, explains Cristina Chiappini of the AIP.

This technique is crucial in preparing for future observations with multi-object spectroscopy, such as 4MIDABLE-LR, a large study of the galactic disk and bulge that will be part of the European Southern Observatory (ESO) 4MOST project. in Chile.

“The new model approach provides detailed maps of the overall chemical composition of the Milky Way, corroborating the distribution of young and old stars. The data shows the concentration of metal-rich stars in the inner regions of the galaxy, including the bar and the bulge, with enormous statistical power,” adds Friedrich Anders from the ICCUB.

The team also used the model to map young, massive hot stars throughout the galaxy, highlighting the distant, poorly studied regions in which stars form. The data also reveals that there are a number of “stellar voids” in our Milky Way – areas that host very few young stars. Additionally, the data demonstrate where the three-dimensional distribution of interstellar dust is still poorly resolved.

As Gaia continues to collect data, the ability of machine learning models to handle large data sets quickly and sustainably makes it an essential tool for future astronomical research.

The success of this approach demonstrates the potential of machine learning to revolutionize big data analysis in astronomy and other scientific fields while promoting more sustainable research practices.