Deblending galaxies with Variational Autoencoders: a joint multi-band, multi-instrument approach

Abstract

Blending of galaxies has a major contribution in the systematic error budget of weak lensing studies, affecting photometric and shape measurements, particularly for ground-based, deep, photometric galaxy surveys, such as the Rubin Observatory Legacy Survey of Space and Time (LSST). Existing deblenders mostly rely on analytic modelling of galaxy profiles and suffer from the lack of flexible yet accurate models. We propose to use generative models based on deep neural networks, namely variational autoencoders (VAE), to learn probabilistic models directly from data. We train a VAE on images of centred, isolated galaxies, which we reuse, as a prior, in a second VAE-like neural network in charge of deblending galaxies. We train our networks on simulated images including six LSST bandpass filters and the visible and near-infrared bands of the Euclid satellite, as our method naturally generalises to multiple bands and can incorporate data from multiple instruments. We obtain median reconstruction errors on ellipticities and $r$-band magnitude between $\pm 0.01$ and $\pm 0.05$ respectively in most cases, and ellipticity multiplicative bias of 1.6% for blended objects in the optimal configuration. We also study the impact of decentring and prove the method to be robust. This method only requires the approximate centre of each target galaxy, but no assumptions about the number of surrounding objects, pointing to an iterative detection/deblending procedure we leave for future work. Finally, we discuss future challenges about training on real data and obtain encouraging results when applying transfer learning. Our code is publicly available on GitHub (https://github.com/LSSTDESC/DeblenderVAE).

arXiv commentMatches version accepted by MNRAS (minor changes, abstract shortened). 17 pages

References in corpus (34)

• LSST: from Science Drivers to Reference Design and Anticipated Data Products
• Wide-Field InfrarRed Survey Telescope-Astrophysics Focused Telescope Assets WFIRST-AFTA 2015 Report
• LSST Science Book, Version 2.0
• A Method for Weak Lensing Observations
• Autoencoding beyond pixels using a learned similarity metric
• Cosmology from cosmic shear power spectra with Subaru Hyper Suprime-Cam first-year data
• The Hyper Suprime-Cam Software Pipeline
• Deep Unsupervised Clustering with Gaussian Mixture Variational Autoencoders
• Weak lensing for precision cosmology
• Dark Energy Survey Year 1 Results: Constraints on Extended Cosmological Models from Galaxy Clustering and Weak Lensing
• CFHTLenS revisited: assessing concordance with Planck including astrophysical systematics
• Practical Weak Lensing Shear Measurement with Metacalibration
• Ladder Variational Autoencoders
• Weak Lensing and Dark Energy
• Dark Synergy: Gravitational Lensing and the CMB
• Weak lensing shear calibration with simulations of the HSC survey
• The Third Gravitational Lensing Accuracy Testing (GREAT3) Challenge Handbook
• CosmoDC2: A Synthetic Sky Catalog for Dark Energy Science with LSST
• Metacalibration: Direct Self-Calibration of Biases in Shear Measurement
• Euclid Imaging Consortium Science Book
• The LSST Dark Energy Science Collaboration (DESC) Science Requirements Document
• SCARLET: Source separation in multi-band images by Constrained Matrix Factorization
• Transfer learning for galaxy morphology from one survey to another
• Variational Deep Embedding: An Unsupervised and Generative Approach to Clustering
• Photometry of high-redshift blended galaxies using deep learning
• Topic-Guided Variational Autoencoders for Text Generation
• The Ellipticity Distribution of Ambiguously Blended Objects
• Deblending galaxy superpositions with branched generative adversarial networks
• The continuous Bernoulli: fixing a pervasive error in variational autoencoders
• A Hybrid Convolutional Variational Autoencoder for Text Generation
• Multi-band morpho-Spectral Component Analysis Deblending Tool (MuSCADeT): Deblending colourful objects
• Autoencoders for music sound modeling: a comparison of linear, shallow, deep, recurrent and variational models
• The LSST DESC Data Challenge 1: Generation and Analysis of Synthetic Images for Next Generation Surveys
• Galaxy shape measurement synergies between LSST and Euclid

Cited by in corpus (31)

• The DAWES review 10: The impact of deep learning for the analysis of galaxy surveys
• Astronomia ex machina: a history, primer, and outlook on neural networks in astronomy
• Deep Generative Models for Galaxy Image Simulations
• Realistic galaxy image simulation via score-based generative models
• Machine Learning for Observational Cosmology
• Detection, Instance Segmentation, and Classification for Astronomical Surveys with Deep Learning (DeepDISC): Detectron2 Implementation and Demonstration with Hyper Suprime-Cam Data
• Euclid preparation: XIII. Forecasts for galaxy morphology with the Euclid Survey using Deep Generative Models
• A Joint Roman Space Telescope and Rubin Observatory Synthetic Wide-Field Imaging Survey
• The PAU survey: Estimating galaxy photometry with deep learning
• DeepSZ: Identification of Sunyaev-Zel'dovich Galaxy Clusters using Deep Learning
• Realistic galaxy images and improved robustness in machine learning tasks from generative modelling
• Deblending Galaxies with Generative Adversarial Networks
• Euclid preparation. LIII. LensMC, weak lensing cosmic shear measurement with forward modelling and Markov Chain Monte Carlo sampling
• Gaussian Process Classification for Galaxy Blend Identification in LSST
• Galaxy Deblending using Residual Dense Neural networks
• Deep Transfer Learning for Blended Source Identification in Galaxy Survey Data
• Score-matching neural networks for improved multi-band source separation
• Neural Network Based Point Spread Function Deconvolution For Astronomical Applications
• Accurately Estimating Redshifts from CSST Slitless Spectroscopic Survey using Deep Learning
• Galaxy Imaging with Generative Models: Insights from a Two-Models Framework
• Joint survey processing: combined resampling and convolution for galaxy modelling and deblending
• Deblending Overlapping Galaxies in DECaLS Using Transformer-Based Algorithm: A Method Combining Multiple Bands and Data Types
• Accelerating lensed quasar discovery and modeling with physics-informed variational autoencoders
• Identification of galaxy shreds in large photometric catalogs using Convolutional Neural Networks
• GalaxyGenius: Mock galaxy image generator for various telescopes from hydrodynamical simulations
• Blending effects on shear measurement synergy between Euclid-like and LSST-like surveys
• A machine learning approach to assessing the presence of substructure in quasar host galaxies using the Hyper Suprime-Cam Subaru Strategic Program
• Investigation on deep learning-based galaxy image translation models
• A Bayesian Convolutional Neural Network for Robust Galaxy Ellipticity Regression
• Comparison of Graphcore IPUs and Nvidia GPUsfor cosmology applications
• Extracting latent representations from X-ray spectra. Classification, regression, and accretion signatures of Chandra sources