James Webb Space Telescope Feed Post

Arxiv: Dust Extinction Measures for z\sim 8 Galaxies using Machine Learning on JWST Imaging Published: 3/27/2024 1:17:51 PM Updated: 3/27/2024 1:17:51 PM

Paper abstract: We present the results of a machine learning study to measure the dustcontent of galaxies observed with JWST at z > 6 through the use of trainedneural networks based on high-resolution IllustrisTNG simulations. Dust is animportant unknown in the evolution and observability of distant galaxies and isdegenerate with other stellar population features through spectral energyfitting. As such, we develop and test a new SED-independent machine learningmethod to predict dust attenuation and sSFR of high redshift (z > 6) galaxies.Simulated galaxies were constructed using the IllustrisTNG model, with avariety of dust contents parameterized by E(B-V) and A(V) values, then used totrain Convolutional Neural Network (CNN) models using supervised learningthrough a regression model. We demonstrate that within the context of thesesimulations, our single and multi-band models are able to predict dust contentof distant galaxies to within a 1\sigma dispersion of A(V) ~ 0.1.Applied to spectroscopically confirmed z > 6 galaxies from the JADES and CEERSprograms, our models predicted attenuation values of A(V) < 0.7 for allsystems, with a low average (A(V) = 0.28). Our CNN predictions show larger dustattenuation but lower amounts of star formation compared to SED fitted values.Both results show that distant galaxies with confirmed spectroscopy are notextremely dusty, although this sample is potentially significantly biased. Wediscuss these issues and present ideas on how to accurately measure dustfeatures at the highest redshifts using a combination of machine learning andSED fitting.