Detecting galactic filaments using machine learning

[ad_1]

Detecting galactic filaments using machine learning

An example of a galactic plane region for superimposing the obtained result. The upper left image shows the region seen in the near infrared emission (K-band, 2MASS scan). This data was not used for training but was used here to empirically validate the score obtained with supervised learning and segmentation (lower left image). This image shows a map of the probability that a pixel belongs to the “fuse” category, which is the structure we were trying to determine from training. The upper right image shows the data used in this study, which shows the plume density distribution (the amount of material on the line of sight) obtained from infrared space satellite Herschel data. Black squares show saturated regions where no material information could be obtained. The lower right image shows threads known before our study, whose structures were used as masks for supervised learning using Unet and Unet++ convolutional networks. credit: Astronomy and astrophysics (2022). DOI: 10.1051/0004-6361/202244103

Star formation in galaxies occurs in filaments made up of gas (mainly hydrogen) and tiny solid particles called interstellar dust, which is mainly carbon. Depending on the location of these filaments and their physical properties (density, temperature) they can be difficult to detect in the data. In particular, low-density filaments or filaments located in very high emissivity regions are generally not detected.

Innovatively and Multidisciplinary approachA team involving some CNRS labs tested the interest of supervised machine learning to try to detect strings It is located on the plane of our galaxy. This approach is based on the current results of filament detection using conventional extraction methods.

The extracted threads are used to train Unet and Unet++ convolutional networks. The trained model learns to recognize filaments and then allows the researchers to create a galaxy-level image where each pixel is represented by its probability (between 0 and 1) of belonging to the acquired filament class.

The results of the learning approach show that this method can detect threads that have not been previously identified by the usual detection methods. New leads are being uncovered and can be confirmed by an experimental approach using available data at other wavelengths not currently used in the learning process.

The results have been published in the journal Astronomy and astrophysics.

The goal of this project, called BigSF, is to study star formation in our galaxy by combining the large amount of available data and machine learning.

more information:
a. Zavagno et al., Supervised Machine Learning on Galactic Filaments, Astronomy and astrophysics (2022). DOI: 10.1051/0004-6361/202244103

the quote: Discovering Galactic Filaments Using Machine Learning (2023, January 23) Retrieved January 23, 2023 from https://phys.org/news/2023-01-galactic-filaments-machine.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.



[ad_2]

Source link

Related Posts

Precaliga