Speaker
Description
In this work we evaluate the performance of several techniques for identifying young stellar objects (YSOs) using photometric data from the WISE and 2MASS surveys. We apply the color–based classification scheme proposed by Koenig & Leisawitz (2014) and compare it with multiple Machine Learning approaches, including K-Nearest Neighbours, Support Vector Classifier, Gradient Boosting, and in particular Random Forest, which is the main focus of our analysis. All classification methods are trained and tested on a sample of 688 155 sources previously labeled as YSO or non-YSO based on the Konkoly Optical YSO Catalogue (Marton et al. 2023). We also perform a comparative analysis across several star-forming regions, including Upper Scorpius, Ophiuchus, Taurus, and Orion, highlighting regional differences in classification performance and YSO demographics.
