Connecting optical morphology, environment, and HI mass fraction for low-redshift galaxies using deep learning

Abstract

A galaxy's morphological features encode details about its gas content, star formation history, and feedback processes, which play important roles in regulating its growth and evolution. We use deep convolutional neural networks (CNNs) to learn a galaxy's optical morphological information in order to estimate its neutral atomic hydrogen (HI) content directly from SDSS $gri$ image cutouts. We are able to accurately predict a galaxy's logarithmic HI mass fraction, $\mathcal{M}\equiv \log (M_{\mathrm{HI}}/M_{\star })$, by training a CNN on galaxies in the ALFALFA 40% sample. Using pattern recognition (PR), we remove galaxies with unreliable $\mathcal{M}$ estimates. We test CNN predictions on the ALFALFA 100%, xGASS, and NIBLES catalogs, and find that the CNN consistently outperforms previous estimators. The HI-morphology connection learned by the CNN appears to be constant in low- to intermediate-density galaxy environments, but it breaks down in the highest-density environments. We also use a visualization algorithm, Gradient-weighted Class Activation Maps (Grad-CAM), to determine which morphological features are associated with low or high gas content. These results demonstrate that CNNs are powerful tools for understanding the connections between optical morphology and other properties, as well as for probing other variables, in a quantitative and interpretable manner.

arXiv comment23 pages, 13 figures, accepted to ApJ. All code public at https://github.com/jwuphysics/HI-convnets

References in corpus (66)

• The NumPy array: a structure for efficient numerical computation
• Decoupled Weight Decay Regularization
• Galactic Stellar and Substellar Initial Mass Function
• Deep Residual Learning for Image Recognition
• The Origin of the Mass--Metallicity Relation: Insights from 53,000 Star-Forming Galaxies in the SDSS
• LSST: from Science Drivers to Reference Design and Anticipated Data Products
• The physical properties of star forming galaxies in the low redshift universe
• A systematic study of the class imbalance problem in convolutional neural networks
• Self-Attention Generative Adversarial Networks
• Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour
• Stellar Masses and Star Formation Histories for 10^5 Galaxies from the Sloan Digital Sky Survey
• Overview of the DESI Legacy Imaging Surveys
• UV Star Formation Rates in the Local Universe
• Wide-Field InfrarRed Survey Telescope-Astrophysics Focused Telescope Assets WFIRST-AFTA 2015 Report
• fastai: A Layered API for Deep Learning
• Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps
• When Does Label Smoothing Help?
• The Arecibo Legacy Fast ALFA Survey: I. Science Goals, Survey Design and Strategy
• A disciplined approach to neural network hyper-parameters: Part 1 -- learning rate, batch size, momentum, and weight decay
• Rotation-invariant convolutional neural networks for galaxy morphology prediction
• The Arecibo Legacy Fast ALFA Survey: The alpha.40 HI Source Catalog, its Characteristics and their Impact on the Derivation of the HI Mass Function
• The Fourteenth Data Release of the Sloan Digital Sky Survey: First Spectroscopic Data from the extended Baryon Oscillation Spectroscopic Survey and from the second phase of the Apache Point Observatory Galactic Evolution Experiment
• Improved background subtraction for the Sloan Digital Sky Survey images
• xCOLD GASS: the complete IRAM-30m legacy survey of molecular gas for galaxy evolution studies
• The Arecibo Legacy Fast ALFA Survey: The ALFALFA Extragalactic HI Source Catalog
• Visualizing and Understanding Convolutional Networks
• The GALEX Arecibo SDSS Survey. I. Gas Fraction Scaling Relations of Massive Galaxies and First Data Release
• Preparing Red-Green-Blue (RGB) Images from CCD Data
• The Atlas3D project -- XIII. Mass and morphology of HI in early-type galaxies as a function of environment
• xGASS: Total cold gas scaling relations and molecular-to-atomic gas ratios of galaxies in the local Universe
• Lookahead Optimizer: k steps forward, 1 step back
• The Atlas3D project -- XXIX. The new look of early-type galaxies and surrounding fields disclosed by extremely deep optical images
• A Cosmic Variance Cookbook
• The Arecibo Legacy Fast ALFA Survey: The Galaxy Population Detected by ALFALFA
• Halpha Morphologies and Environmental Effects in Virgo Cluster Spiral Galaxies
• The DEEP2 Galaxy Redshift Survey: The Role of Galaxy Environment in the Cosmic Star-Formation History
• Cold gas stripping in satellite galaxies: from pairs to clusters
• The GALEX Arecibo SDSS Survey. VIII. Final Data Release -- The Effect of Group Environment on the Gas Content of Massive Galaxies
• Photometric redshifts from SDSS images using a Convolutional Neural Network
• Linking Gas Fractions to Bimodalities in Galaxy Properties
• Bag of Tricks for Image Classification with Convolutional Neural Networks
• Searching for pulsars using image pattern recognition
• Atomic hydrogen in IllustrisTNG galaxies: the impact of environment parallelled with local 21-cm surveys
• Estimating the H I gas fractions of galaxies in the local Universe
• Transfer learning for galaxy morphology from one survey to another
• The Hubble Sequence at $z\sim 0$ in the IllustrisTNG simulation with deep learning
• DeepCMB: Lensing Reconstruction of the Cosmic Microwave Background with Deep Neural Networks
• ALFALFA HI Data Stacking III. Comparison of environmental trends in HI gas mass fraction and specific star formation rate
• Data-Driven Reconstruction of Gravitationally Lensed Galaxies using Recurrent Inference Machines
• The clustering of galaxies as a function of their photometrically-estimated atomic gas content
• ASKAP HI imaging of the galaxy group IC 1459
• The HI Content of Galaxies in Groups and Clusters as Measured by ALFALFA
• Environmental dependence of the HI mass function in the ALFALFA 70% catalogue
• NIBLES - an HI census of stellar mass selected SDSS galaxies: I. The Nançay HI survey
• Deep Learning at Scale for the Construction of Galaxy Catalogs in the Dark Energy Survey
• RESOLVE Survey Photometry and Volume-limited Calibration of the Photometric Gas Fractions Technique
• Using convolutional neural networks to predict galaxy metallicity from three-color images
• The RESOLVE Survey Atomic Gas Census and Environmental Influences on Galaxy Gas Reservoirs
• On the Classification of UGC1382 as a Giant Low Surface Brightness Galaxy
• GASS 3505: the prototype of HI-excess, passive galaxies
• The infra-red luminosities of ~332,000 SDSS galaxies predicted from artificial neural networks and the Herschel Stripe 82 survey
• Do Androids Dream of Magnetic Fields? Using Neural Networks to Interpret the Turbulent Interstellar Medium
• Predicting the Neutral Hydrogen Content of Galaxies From Optical Data Using Machine Learning
• Pattern recognition in the ALFALFA.70 and Sloan Digital Sky Surveys: A catalog of $\sim$ 500,000 HI gas fraction estimates based on artificial neural networks
• The GALEX Arecibo SDSS Survey VII: The Bivariate Neutral Hydrogen-Stellar Mass Function for Massive Galaxies
• Classifying galaxies according to their HI content

Cited by in corpus (10)

• The SAGA Survey. II. Building a Statistical Sample of Satellite Systems around Milky Way-like Galaxies
• HOLISMOKES -- IX. Neural network inference of strong-lens parameters and uncertainties from ground-based images
• HOLISMOKES -- X. Comparison between neural network and semi-automated traditional modeling of strong lenses
• Deep Learning Assessment of galaxy morphology in S-PLUS DataRelease 1
• How the Galaxy-Halo Connection Depends on Large-Scale Environment
• Predicting galaxy spectra from images with hybrid convolutional neural networks
• Insights on Galaxy Evolution from Interpretable Sparse Feature Networks
• Radio Galaxy Zoo: Morphological classification by Fanaroff-Riley designation using self-supervised pre-training
• Extending the SAGA Survey (xSAGA) I: Satellite Radial Profiles as a Function of Host Galaxy Properties
• A machine learning approach to estimating HI deficiency in galaxies