ExoDNN: Enhancing exoplanet detection with artificial intelligence. Application to Gaia Data Release 3

Applications of AI


ExoDNN: Enhancing exoplanet detection with artificial intelligence. Application to Gaia Data Release 3

Relationship diagram between the color obtained by subtracting RP from Gaia BP and the absolute magnitude of the Gaia DR3 catalog. The 7414 final candidates reported by ExoDNN are shown in green after filtering. Predictions are also shown for several reference objects, including Proxima Centauri, LHS1610, BD+05 5128, HD164604, Gaia 3b, Gaia 4b, and Gaia 5b. Sources with known companions detected by ExoDNN are marked with green boxes, while red crosses refer to reference objects with the model predicted probability ^p < p0, i.e., undetected. The new candidate closest to the ExoDNN is highlighted with a blue circle to the left of the main sequence. -- astro-ph.EP

We combine Gaia Data Release 3 and artificial intelligence to enhance current statistics for stellar companions, especially within regions of orbital period versus mass parameter space that are poorly constrained by radial velocity and transit detection methods.

Use supervised learning to train a deep neural network to recognize characteristic distributions of fit quality statistics corresponding to non-single-star Gaia DR3 astronomical analysis solutions.

Based on these fit quality statistics, we generate a deep learning model, ExoDNN, that predicts the probability that a DR3 source hosts an unresolved companion. We applied ExoDNN’s predictive capabilities to a limited sample of F, G, K, and M stars in Gaia DR3 to generate a list of 7414 candidate stars hosting companion stars.

The stellar properties of these candidates, such as mass and metallicity, are similar to those of the Gaia DR3 non-single star sample. We also identify synergies with future observatories such as PLATO and propose a follow-up strategy aimed at investigating the most promising candidates among those samples.

A. Abreu, J. Lilobox, AM Perez-Garcia, J. Saalman, JHJ De Bruyne, C. Cifuentes

Comments: 15 pages, 13 figures
Subject: Earth and Planetary Astrophysics (astro-ph.EP). Instruments and Methods of Astrophysics (astro-ph.IM); Solar and Stellar Astrophysics (astro-ph.SR)
Quote: arXiv:2602.02910 [astro-ph.EP] (or arXiv:2602.02910v1 [astro-ph.EP] for this version)
https://doi.org/10.48550/arXiv.2602.02910
focus to learn more
Related DOI:
https://doi.org/10.1051/0004-6361/202555598
focus to learn more
Post history
Birthplace: Asier Abreu
[v1] Monday, February 2, 2026 23:36:24 UTC (14,816 KB)
https://arxiv.org/abs/2602.02910

Astrobiology, astronomy, stellar cartography,



Source link