AstroAI Lunch Talks - October 20, 2025 - Mykyta Kliapets
20 Oct 2025 - Joshua Wing
The video can be found here: https://www.youtube.com/watch?v=XoZcnIHntv8
Speaker: Mykyta Kliapets (MIT)
Title: Hunting BAF-type Hybrid Pulsators in TESS with Positive Unlabelled Learning
Abstract: Hybrid (γ Doradus / δ Scuti and SPB / β Cephei) stars are perfect laboratories for probing stellar physics of intermediate and high-mass stars. Exhibiting both pressure (p) modes sensitive to outer interior layers and gravity (g) modes sensitive to near-core regions, they allow to probe outer and inner stellar interior layers simultaneously. While hundreds of p- and g-mode pulsators discovered by Kepler and the Transiting Exoplanet Survey Satellite (TESS) later revealed to be hybrids, many remain to be discovered in TESS data. However, manual vetting of all observations made by TESS is impractical. This is where machine learning comes into play, allowing to efficiently analyze large astronomical datasets and flag candidate pulsators for asteroseimic follow-up exploitations. In this study, we present our pipeline utilizing a periodogram smart binning method for feature extraction and a classifier from a semi-supervised machine learning paradigm, which is currently largely under-utilized in variability studies. We conclude with an overview of global stellar parameters and rotational properties of our candidate hybrids from TESS.