Study of the polarised emission of the interstellar medium of the Magellanic Clouds using big data analysis and machine learning

Grant funding 2020-2021
Ministry of Education and Sciences

The aim of the project is to develop a neural network model to identify filamentart structures in interstellar space, since a detailed study of filamentary structures in the Milky Way can provide information about the structure of the magnetic field. In turn, this will allow us to remove the contributtion from the interstellar dust of the Milky Way in the Magellanic Cloud itself.

Currently, due to the emergence of a large amount of scientific data and the increased computational costs for research, there is a need for their most efficient analysis and processing. The use of machine learning makes it possible to speed up data analysis with lower computational costs, as well as bypass the computational problems that exist in classical methods. At the moment, there is no detailed study of polarized radiation in the Magellanic clouds, while the experimental base, theoretical and empirical methods have been developed.

The expected results are a neural network model and maps of polarized emission of interstellar dust in the Magellanic Cloud. By the end of 2021, training data samples have been built, an initial neural network has been designed, and a method for eliminating foreground radiation has been developed.