Marchant Cortés, Pamela
Chile Universidad de La Serena
NIR Analysis of Milky Way X-ray Sources in the Zone of Avoidance using Machine Learning Tools
Automated methods for classifying extragalactic objects in extensive surveys offer significant advantages over manual approaches in terms of efficiency and consistency. However, the presence of dust extinction near the Galactic disc poses additional challenges. Few studies have explored the application of machine learning (ML) in the Zone of Avoidance (ZoA), characterized by high extinction levels, star crowding, and limited data and studies. One recent study (Daza et al., 2023) employed ML to classify Galaxy and non-Galaxy objects in the VVV and VVVx surveys. Another study (Zhang et al., 2021) used ML to classify STAR/GALAXY/QSO in the 4XMM-DR9 survey, which includes the ZoA.
In this research, we investigate the challenges and benefits of using ML tools for galaxy classification in the ZoA and explore the implications of environmental factors on classification results and their reliability. We address the hypothesis that the analysis area should have conditions similar to the training set, which is not the case in the ZoA. Our analysis reveals significant differences between the sample galaxies and those found throughout the Galactic disc, mainly due to the lack of information on galaxies in the Galactic plane in the training set. Some chosen regions of interest within the ZoA exhibit a high probability of being a galaxy in X-ray data but closely resemble extended Galactic objects.
Our findings emphasize the complexities of using ML for galaxy classification in the ZoA and underscore the importance of considering environmental factors and data distribution to enhance the reliability and accuracy of future studies in this challenging region.