Large Astronomical Time Series Pre-processing and Visualization for Classification using Artificial Neural Networks
David Andrešič, Petr Šaloun, Bronislava Suchanova · 2019
Time series analysis is a growing issue in multiple fields of science. One of the most common task is looking for hidden periods in time series data sets. In this work, we have chosen two large astronomical time series collections from BRITE and Kepler K2 projects and analyzed possible approaches for hidden periods search and their classification. Since these data sets are generally large, we were looking for some automated solution based on artificial neural networks that requires some data pre-processing. This work therefore brings a brief overview of possible solutions for looking for hidden periods in astronomical time series with use of pure artificial neural networks or together with more conventional statistical approaches - mainly from a data pre-processing and its visualization point of view.