Distance determination of molecular clouds in the 1st quadrant of the Galactic plane using deep learning : I. Method and Results

Machine learning has been successfully applied in varied field but whether it is a viable tool for determining the distance to molecular clouds in the Galaxy is an open question. In the Galaxy, the kinematic distance is commonly employed as the distance to a molecular cloud. However, there is a problem in that for the inner Galaxy, two different solutions, the ``Near'' solution, and the ``Far'' solution, can be derived simultaneously. We attempted to construct a two-class (``Near'' or ``Far'') inference model using a Convolutional Neural Network (CNN), a form of deep learning that can capture spatial features generally. In this study, we used the CO dataset toward the 1st quadrant of the Galactic plane obtained with the Nobeyama 45-m radio telescope (l = 62-10 degree, |b| < 1 degree). In the model, we applied the three-dimensional distribution (position-position-velocity) of the 12CO (J=1-0) emissions as the main input. The dataset with ``Near'' or ``Far'' annotation was made from the HII region catalog of the infrared astronomy satellite WISE to train the model. As a result, we could construct a CNN model with a 76% accuracy rate on the training dataset. By using the model, we determined the distance to molecular clouds identified by the CLUMPFIND algorithm. We found that the mass of the molecular clouds with a distance of < 8.15 kpc identified in the 12CO data follows a power-law distribution with an index of about -2.3 in the mass range of M >10^3 Msun. Also, the detailed molecular gas distribution of the Galaxy as seen from the Galactic North pole was determined.