Hostname: page-component-7b7567c8c-m6dxt Total loading time: 0 Render date: 2026-11-19T03:10:31.933Z Has data issue: false hasContentIssue false

Application of Machine Learning Methods to the Identification of Star-Forming Regions

Published online by Cambridge University Press:  05 October 2026

K. V. Plakitina*
Affiliation:
Institute of Astronomy of the Russian Academy of Sciences
M. S. Kirsanova
Affiliation:
Institute of Astronomy of the Russian Academy of Sciences
Rights & Permissions [Opens in a new window]

Abstract

Modern astronomy handles vast amounts of spectral data obtained from observations of molecular emissions across different sources, such as dark infrared clouds, protostars, and HII regions. Traditional analytical methods become increasingly ineffective when dealing with such large datasets. In this study, we investigate the use of machine learning techniques to analyse molecular emission lines and classify molecular clumps. Using data from the MALT90 survey, which observes the Galactic plane at 3 mm wavelengths, we explore various clustering algorithms and dimensionality reduction techniques to simplify the process of object identification. Our results show that machine learning approaches successfully distinguish molecular clumps with active star formation from other types of astronomical objects. This approach holds promise for processing data collected with advanced facilities like JWST and ALMA, leading to more efficient analysis of astronomical data.

Information

Type
Contributed Paper
Copyright
© The Author(s), 2026. Published by Cambridge University Press on behalf of The International Astronomical Union
Figure 0

Figure 1.Distribution of clump categories in the two most stable HDBSCAN clusters from t-SNE on five molecular lines (HCO+$\mathrm{HCO}^{+}$, HNC, N2H+${\mathrm{N}_{2}\mathrm{H}^{+}}$, HCN, C2H$\mathrm{C}_{2}\mathrm{H}$). Bars show MALT90 classification (Q, A, C, H, P, U); percentages give fractions and numbers show counts. Left panel: before classification of uncertain clumps. Right panel: after classification.

Figure 1

Figure 2.Distribution of clump classes in the two most stable HDBSCAN clusters from t-SNE on sixteen molecular lines (HCO+$\mathrm{HCO}^{+}$, HNC, N2H+${\mathrm{N}_{2}\mathrm{H}^{+}}$, HCN, C2H$\mathrm{C}_{2}\mathrm{H}$, H13CO+$\mathrm{H}^{1}{3}\mathrm{CO}^{+}$, HN13C$\mathrm{HN}^{1}{3}\mathrm{C}$,13CS$^{1}3\mathrm{CS}$,13C34S$^{1}3\mathrm{C}^{3}4\mathrm{S}$, HC13CCN$\mathrm{HC}^{1}3\mathrm{CCN}$, HNCO, CH3CN$\mathrm{CH}_{3}\mathrm{CN}$, HC3N$\mathrm{HC}_{3}\mathrm{N}$, HNCO, H 41α$\alpha$. Left panel: before classification of uncertain clumps. Right panel: after classification.