Aprendizaje automático para identificar poblaciones estelares en galaxias cercanas

GUSTAVO BAUME; MARIA JIMENA RODRIGUEZ; CARLOS FEINSTEIN; ERIKA GULARTE

Rosario

Congreso; 62ª Reunión Anual de la Asociación Argentina de Astronomía; 2020

A study of different stellar populations in nearby galaxies has been carried out. This has been based on multi-band photometric data obtained with the Hubble Space Telescope. In the analysis, unsupervised machine learning techniques have been applied in order to recognize both the stellar populations and the groups of stars in the youngest population. In both cases, different clustering algorithms have been used and their efficiency has been evaluated. The applied methodology has allowed to carry out the task without the need for preconceived criteria. Additionally, it has been possible to characterize the spatial distribution of each of the stellar populations considering their similarities with a fractal-type structure. In this way, it has been possible to identify the youngest populations with a hierarchical structure and the more evolved populations with homogeneous distributions, except for fluctuations on a very large scale.