Speaker
Description
In astronomy, transients are a source of new discoveries about the universe. Kilonovae are a type of transient resulting from the collision between neutron stars or between a neutron star and a black hole in compact binary systems, and currently, it is the only object with electromagnetic and gravitational wave counterparts (GW170817). As we navigate in the Big Data era, and expect the intensification of it with observatories such as LSST generating several alerts of transients per night, the use of Machine Learning algorithms enables an effective response for their analysis. This study seeks to understand how the use of Deep Learning can help in the identification of Kilonovae's spectral energy distribution (SED), testing different networks and studying which model achieves the best results when trying to differ between Supernovae e Kilonovae by its spectrum. The initial results indicate the accuracy of the recurrent neural network (RNN) architecture designed to classify the spectra, and with that in mind we can consider the possibility of future application of this methodology as a follow-up for the O4 LIGO-Virgo-KAGRA observing run, using data from observatories such as LSST to the classification of these transients searching for multi-messenger data.
