Research Division Seminar
The Kepler Pixel Project and variable star classification with 'computer vision'

Abstract

Kepler photometry was so precise that new ways could be developed to harvest the great wealth of quasi-continuous data that has never been accessible from the ground. We initiated a project that we dubbed The Kepler Pixel Project in order to explore approaches and to discover new pulsating stars and other time-variable objects. During the project we examined individual pixels of the original Kepler mission to find interesting objects around the main Kepler targets. Specifically we launched a subproject to find background, faint RR Lyrae stars that are missing from the original Kepler sample. Altogether we found 26 new RR Lyrae stars, increasing the Kepler original RR Lyrae sample by 50%. In this talk I'll present the latest results of this project. In addition to RR Lyrae stars I will also show results on ~1000 new eclipsing binaries found in the framework of the same project.

Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST) is one of the most important ground-based astronomy projects of the coming decade. In the second half of this talk I will present my research group's work on classification of variable stars with machine learning methods which is part of the Hungarian in-kind LSST contribution. The novelty of our method is that we use images of light curves, such as a human classifier would do. The method gives surprisingly good results based on the shape of light curves only, but can be further improved if additional astrophysical parameters (distance, amplitude, colors, etc.) are taken into account.