Joana C. Leite
DCC-FCUP

Abstract
Recent releases of large-scale astronomical surveys are generating more data than ever, increasing the demand for new accurate and automatic methods to categorize these data and extract information without the need of persistent human intervention. Specifically, deriving the atmospheric parameters and chemical abundances of certain elements in stars plays a pivotal role in elucidating their evolutionary stages and offers valuable information on revealing the history of galaxies. These are also relevant in exoplanet detection and research, as the composition of an exoplanet is related to the composition of the host star and its physical properties.
Estimating these parameters and abundances remains, nevertheless, a cumbersome task that, despite some automation, still requires some degree of manual inspection. Unconventional data-driven techniques hold the promise of efficiently dealing with vast collections of data while still rendering results of astrophysical value.
In this talk, I will introduce a multi-target machine learning strategy for an accurate simultaneous estimation of stellar atmospheric parameters (Effective Temperature, Surface gravity and metallicity) and 13 chemical abundances (CuI, ZnI, SrI, YII, ZrII, BaII, CeII, AlI, MgI, SiI, CaI, TiI, TiII) for FGK stars using information from stellar spectra. I will briefly cover data statistical analysis, discuss models' and dimensionality reduction, and delve into the possibility of interpretability.

2024 June 19, 13:30

IA/U.Porto
Centro de Astrofísica da Universidade do Porto (Classroom)
Rua das Estrelas, 4150-762 Porto