Volume 633, January 2020
|Number of page(s)||17|
|Published online||23 January 2020|
Deriving star cluster parameters with convolutional neural networks
II. Extinction and cluster-background classification
Vilnius University Observatory, Saulėtekio av. 3, 10257 Vilnius, Lithuania
e-mail: firstname.lastname@example.org, email@example.com
2 Center for Physical Sciences and Technology, Saulėtekio av. 3, 10257 Vilnius, Lithuania
Accepted: 19 November 2019
Context. Convolutional neural networks (CNNs) have been established as the go-to method for fast object detection and classification of natural images. This opens the door for astrophysical parameter inference on the exponentially increasing amount of sky survey data. Until now, star cluster analysis was based on integral or resolved stellar photometry, which limit the amount of information that can be extracted from individual pixels of cluster images.
Aims. We aim to create a CNN capable of inferring star cluster evolutionary, structural, and environmental parameters from multiband images and to demonstrate its capabilities in discriminating genuine clusters from galactic stellar backgrounds.
Methods. A CNN based on the deep residual network (ResNet) architecture was created and trained to infer cluster ages, masses, sizes, and extinctions with respect to the degeneracies between them. Mock clusters placed on M 83 Hubble Space Telescope images utilizing three photometric passbands (F336W, F438W, and F814W) were used. The CNN is also capable of predicting the likelihood of the presence of a cluster in an image and quantifying its visibility (S/N).
Results. The CNN was tested on mock images of artificial clusters and has demonstrated reliable inference results for clusters of ages ≲100 Myr, extinctions AV between 0 and 3 mag, masses between 3 × 103 and 3 × 105 M⊙, and sizes between 0.04 and 0.4 arcsec at the distance of the M 83 galaxy. Real M 83 galaxy cluster parameter inference tests were performed with objects taken from previous studies and have demonstrated consistent results.
Key words: methods: data analysis / methods: statistical / techniques: image processing / galaxies: individual: M 83 / galaxies: clusters: general
© ESO 2020
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.