Pursuing variable stars with the power of machine learning
Attila Bódi
Konkoly Observatory, Budapest, Hungary


In this presentation I will briefly summarize the main results and ongoing projects of the Konkoly machine learning (ML) group. The group was formed in September 2019 with the aim of getting to know ML techniques and applying them to different astronomical tasks. Because the topic is intrinsically diverse, my talk will cover a number of astronomical fields. We have achieved successful results in the detection of flares of active low-mass stars in Kepler and TESS observations. We have used both supervised and unsupervised methods to classify non-pulsating and pulsating variable stars observed by the ground-based Optical Gravitational Lensing Experiment. Our current projects aim to achieve more ambitious outcomes. With the advent of long-time-span and large-scale surveys, the amount of data increases so dramatically that astronomers won't be able to handle those as they are used to, manually. Therefore, we are working on hot topics such as classifying variable stars that will be observed by the Vera C. Rubin Observatory and looking for needles in a hay-stack, a.k.a. finding anomalies in big data.

Here you can play back the presentation