New variable Stars from the Photographic Archive: Semi-automated Discoveries, Attempts of Automatic Classification, and the New Field 104 Her

TL;DR

This study reports the discovery of 275 new variable stars from photographic archives, explores automated classification using Random Forest, and highlights the potential for large-scale sky surveys to identify many more variables.

Contribution

It introduces a semi-automated approach to discover and classify variable stars from photographic data, demonstrating the feasibility of automated classification with noisy data.

Findings

275 new variable stars discovered and cataloged

Automated classification with Random Forest is feasible on photographic data

Further improvements needed for automated techniques to handle large survey data

Abstract

Using 172 plates taken with the 40-cm astrograph of the Sternberg Astronomical Institute (Lomonosov Moscow University) in 1976-1994 and digitized with the resolution of 2400 dpi, we discovered and studied 275 new variable stars. We present the list of our new variables with all necessary information concerning their brightness variations. As in our earlier studies, the new discoveries show a rather large number of high-amplitude Delta Scuti variables, predicting that many stars of this type remain not detected in the whole sky. We also performed automated classification of the newly discovered variable stars based on the Random Forest algorithm. The results of the automated classification were compared to traditional classification and showed that automated classification was possible even with noisy photographic data. However, further improvement of automated techniques is needed, which is especially important having in mind the very large numbers of new discoveries expected from all-sky surveys.