Realistic galaxy images and improved robustness in machine learning tasks from generative modelling

Abstract

We examine the capability of generative models to produce realistic galaxy images. We show that mixing generated data with the original data improves the robustness in downstream machine learning tasks. We focus on three different data sets; analytical Sérsic profiles, real galaxies from the COSMOS survey, and galaxy images produced with the SKIRT code, from the IllustrisTNG simulation. We quantify the performance of each generative model using the Wasserstein distance between the distributions of morphological properties (e.g. the Gini-coefficient, the asymmetry, and ellipticity), the surface brightness distribution on various scales (as encoded by the power-spectrum), the bulge statistic and the colour for the generated and source data sets. With an average Wasserstein distance (Fréchet Inception Distance) of $7.19\times 10^{-2}(0.55)$, $5.98\times 10^{-2}(1.45)$ and $5.08\times 10^{-2}(7.76)$ for the Sérsic, COSMOS and SKIRT data set, respectively, our best models convincingly reproduce even the most complicated galaxy properties and create images that are visually indistinguishable from the source data. We demonstrate that by supplementing the training data set with generated data, it is possible to significantly improve the robustness against domain-shifts and out-of-distribution data. In particular, we train a convolutional neural network to denoise a data set of mock observations. By mixing generated images into the original training data, we obtain an improvement of $11$ and $45$ per cent in the model performance regarding domain-shifts in the physical pixel size and background noise level, respectively.

arXiv comment33 pages, 21 figures, submitted to MNRAS, comments welcome

References in corpus (31)

• SciPy 1.0--Fundamental Algorithms for Scientific Computing in Python
• Array Programming with NumPy
• Astropy: A Community Python Package for Astronomy
• The Sloan Digital Sky Survey: Technical Summary
• The Astropy Project: Building an inclusive, open-science project and status of the v2.0 core package
• An Introduction to Variational Autoencoders
• Deep Visual Domain Adaptation: A Survey
• The Cosmic Evolution Survey (COSMOS) -- Overview
• First results from the IllustrisTNG simulations: matter and galaxy clustering
• First results from the IllustrisTNG simulations: the stellar mass content of groups and clusters of galaxies
• First results from the IllustrisTNG simulations: the galaxy color bimodality
• First results from the IllustrisTNG simulations: A tale of two elements -- chemical evolution of magnesium and europium
• First results from the IllustrisTNG simulations: radio haloes and magnetic fields
• Skillful Precipitation Nowcasting using Deep Generative Models of Radar
• The Relationship Between Stellar Light Distributions of Galaxies and their Formation Histories
• A New Non-Parametric Approach to Galaxy Morphological Classification
• Galaxy morphology to I=25 mag in the Hubble Deep Field
• The optical morphologies of galaxies in the IllustrisTNG simulation: a comparison to Pan-STARRS observations
• Structural and Photometric Classification of Galaxies - I. Calibration Based on a Nearby Galaxy Sample
• Efficient 3D NLTE dust radiative transfer with SKIRT
• A catalog of visual-like morphologies in the 5 CANDELS fields using deep-learning
• THE CANADA-FRANCE REDSHIFT SURVEY IX: HST Imaging of High-Redshift Field Galaxies
• Star-galaxy Classification Using Deep Convolutional Neural Networks
• Galaxy Morphology and Star Formation in the Illustris Simulation at z=0
• Precision simulation of ground-based lensing data using observations from space
• Deep Generative Models for Galaxy Image Simulations
• Deblending galaxies with Variational Autoencoders: a joint multi-band, multi-instrument approach
• Realistic galaxy image simulation via score-based generative models
• Forging new worlds: high-resolution synthetic galaxies with chained generative adversarial networks
• Weak-lensing shear measurement with machine learning: teaching artificial neural networks about feature noise
• Euclid preparation: XIII. Forecasts for galaxy morphology with the Euclid Survey using Deep Generative Models

Cited by in corpus (1)

• Astronomia ex machina: a history, primer, and outlook on neural networks in astronomy