Generative deep fields: arbitrarily sized, random synthetic astronomical images through deep learning

Abstract

Generative Adversarial Networks (GANs) are a class of artificial neural network that can produce realistic, but artificial, images that resemble those in a training set. In typical GAN architectures these images are small, but a variant known as Spatial-GANs (SGANs) can generate arbitrarily large images, provided training images exhibit some level of periodicity. Deep extragalactic imaging surveys meet this criteria due to the cosmological tenet of isotropy. Here we train an SGAN to generate images resembling the iconic Hubble Space Telescope eXtreme Deep Field (XDF). We show that the properties of 'galaxies' in generated images have a high level of fidelity with galaxies in the real XDF in terms of abundance, morphology, magnitude distributions and colours. As a demonstration we have generated a 7.6-billion pixel 'generative deep field' spanning 1.45 degrees. The technique can be generalised to any appropriate imaging training set, offering a new purely data-driven approach for producing realistic mock surveys and synthetic data at scale, in astrophysics and beyond.

arXiv commentSubmitted to MNRAS. Comments welcome. Code available at https://github.com/Smith42/XDF-GAN and 7.6-billion pixel GDF viewable at http://star.herts.ac.uk/~jgeach/gdf

References in corpus (14)

• The EAGLE project: Simulating the evolution and assembly of galaxies and their environments
• Introducing the Illustris Project: Simulating the coevolution of dark and visible matter in the Universe
• The Cosmic Evolution Survey (COSMOS) -- Overview
• The Hubble Deep Field: Observations, Data Reduction, and Galaxy Photometry
• The HST eXtreme Deep Field XDF: Combining all ACS and WFC3/IR Data on the HUDF Region into the Deepest Field Ever
• Sunrise: Polychromatic Dust Radiative Transfer in Arbitrary Geometries
• Generative Adversarial Networks recover features in astrophysical images of galaxies beyond the deconvolution limit
• Optical colours and spectral indices of $z=0.1$ EAGLE galaxies with 3D dust radiative transfer code SKIRT
• Mock 2dF and SDSS galaxy redshift surveys
• A Virtual Sky with Extragalactic HI and CO Lines for the SKA and ALMA
• Massive Close Pairs Measure Rapid Galaxy Assembly in Mergers at High Redshift
• Precision simulation of ground-based lensing data using observations from space
• Simulation of Astronomical Images from Optical Survey Telescopes using a Comprehensive Photon Monte Carlo Approach
• Astronomical Image Simulation for Telescope and Survey Development

Cited by in corpus (13)

• The DAWES review 10: The impact of deep learning for the analysis of galaxy surveys
• Morpheus: A Deep Learning Framework For Pixel-Level Analysis of Astronomical Image Data
• Astronomia ex machina: a history, primer, and outlook on neural networks in astronomy
• Deep Generative Models for Galaxy Image Simulations
• AstroVaDEr: Astronomical Variational Deep Embedder for Unsupervised Morphological Classification of Galaxies and Synthetic Image Generation
• Realistic galaxy image simulation via score-based generative models
• Painting halos from cosmic density fields of dark matter with physically motivated neural networks
• Investigating Cosmological GAN Emulators Using Latent Space Interpolation
• Deep-CEE I: Fishing for Galaxy Clusters with Deep Neural Nets
• Survey2Survey: A deep learning generative model approach for cross-survey image mapping
• Machine learning technique for morphological classification of galaxies from SDSS. II. The image-based morphological catalogs of galaxies at 0.02<z<0.1
• Using a Conditional Generative Adversarial Network to Control the Statistical Characteristics of Generated Images for IACT Data Analysis
• Generative Adversarial Networks for Astronomical Images Generation