Detection of Carbon stars using statistical Learning -- Seminario de 11am a 12:20pm

"Carbon stars are a unique and not common type of variable stars with an effective temperature 3000 K. They contain more carbon than oxygen in their atmospheres, due to a relatively rare dredging-up effect. Most carbon stars are late red giants and belong to the Asymptotic Giant Branch, which is one of the final stages of stellar evolution. They have roughly the same mean absolute magnitude. These stars are known to spread chemical elements, particularly carbon, to the interstellar medium through eruptions or stellar winds. This makes them useful for a variety of purposes, such as indicating age and distance and aiding in the study of interstellar dust and stellar evolution processes.
Because carbon stars are not common a semi-supervised Machine Learning algorithms have been developed to detect this stars using a very few amount of already known carbon stars, comparing them to a huge set of stars (unlabeled) in order to detect new Galactic carbon stars.

Although there are much less Carbon stars than no Carbon stars, today there are enough data to use supervised Machine Learning models to detect and differentiate them from other stars using a simple, efficient, effective, and interpretable model that can handle a highly unbalanced data."