HOLISMOKES -- XI. Evaluation of supervised neural networks for strong-lens searches in ground-based imaging surveys

Abstract

While supervised neural networks have become state of the art for identifying the rare strong gravitational lenses from large imaging data sets, their selection remains significantly affected by the large number and diversity of nonlens contaminants. This work evaluates and compares systematically the performance of neural networks in order to move towards a rapid selection of galaxy-scale strong lenses with minimal human input in the era of deep, wide-scale surveys. We used multiband images from PDR2 of the HSC Wide survey to build test sets mimicking an actual classification experiment, with 189 strong lenses previously found over the HSC footprint and 70,910 nonlens galaxies in COSMOS. Multiple networks were trained on different sets of realistic strong-lens simulations and nonlens galaxies, with various architectures and data pre-processing. The overall performances strongly depend on the construction of the ground-truth training data and they typically, but not systematically, improve using our baseline residual network architecture. Improvements are found when applying random shifts to the image centroids and square root stretches to the pixel values, adding z band, or using random viewpoints of the original images, but not when adding difference images to subtract emission from the central galaxy. The most significant gain is obtained with committees of networks trained on different data sets, and showing a moderate overlap between populations of false positives. Nearly-perfect invariance to image quality can be achieved by training networks either with large number of bands, or jointly with the PSF and science frames. Overall, we show the possibility to reach a TPR0 as high as 60% for the test sets under consideration, which opens promising perspectives for pure selection of strong lenses without human input using the Rubin Observatory and other forthcoming ground-based surveys.

arXiv comment21 pages, 10 figures, submitted to A&A, comments are welcome

References in corpus (79)

• SciPy 1.0--Fundamental Algorithms for Scientific Computing in Python
• Array Programming with NumPy
• Planck 2015 results. XIII. Cosmological parameters
• Binary companions of evolved stars in APOGEE DR14: Search method and catalog of ~5,000 companions
• LSST: from Science Drivers to Reference Design and Anticipated Data Products
• H0LiCOW XIII. A 2.4% measurement of $H_0$ from lensed quasars: $5.3\sigma$ tension between early and late-Universe probes
• The Hyper Suprime-Cam SSP Survey: Overview and Survey Design
• The Hubble Ultra Deep Field
• Rotation-invariant convolutional neural networks for galaxy morphology prediction
• Galaxy Zoo 2: detailed morphological classifications for 304,122 galaxies from the Sloan Digital Sky Survey
• First Data Release of the Hyper Suprime-Cam Subaru Strategic Program
• The Hyper Suprime-Cam Software Pipeline
• H0LiCOW V. New COSMOGRAIL time delays of HE0435-1223: $H_0$ to 3.8% precision from strong lensing in a flat $\Lambda$CDM model
• The Sloan Lens ACS Survey. V. The Full ACS Strong-Lens Sample
• A catalog of 132,684 clusters of galaxies identified from Sloan Digital Sky Survey III
• The population of galaxy-galaxy strong lenses in forthcoming optical imaging surveys
• Cosmology from the Chinese Space Station Optical Survey (CSS-OS)
• Fast Automated Analysis of Strong Gravitational Lenses with Convolutional Neural Networks
• Galaxy Zoo DECaLS: Detailed Visual Morphology Measurements from Volunteers and Deep Learning for 314,000 Galaxies
• CMU DeepLens: Deep Learning For Automatic Image-based Galaxy-Galaxy Strong Lens Finding
• The BOSS Emission-Line Lens Survey (BELLS). I. A large spectroscopically selected sample of Lens Galaxies at redshift ~ 0.5
• The Halos of Satellite Galaxies: the Companion of the Massive Elliptical Lens SL2S J08544-0121
• Finding Strong Gravitational Lenses in the Kilo Degree Survey with Convolutional Neural Networks
• The MUSE Hubble Ultra Deep Field Survey: II. Spectroscopic redshifts and comparisons to color selections of high-redshift galaxies
• Disentangling Baryons and Dark Matter in the Spiral Gravitational Lens B1933+503
• Photometric redshift estimation via deep learning
• The SDSS-IV extended Baryon Oscillation Spectroscopic Survey: Baryon Acoustic Oscillations at redshift of 0.72 with the DR14 Luminous Red Galaxy Sample
• Photometric Redshift with Bayesian Priors on Physical Properties of Galaxies
• The Strong Gravitational Lens Finding Challenge
• First catalog of strong lens candidates in the COSMOS field
• Classifying Radio Galaxies with Convolutional Neural Network
• Finding strong lenses in CFHTLS using convolutional neural networks
• RingFinder: automated detection of galaxy-scale gravitational lenses in ground-based multi-filter imaging data
• Space Warps: I. Crowd-sourcing the Discovery of Gravitational Lenses
• Space Warps II. New Gravitational Lens Candidates from the CFHTLS Discovered through Citizen Science
• Survey of Gravitationally-lensed Objects in HSC Imaging (SuGOHI). I. Automatic search for galaxy-scale strong lenses
• Finding high-redshift strong lenses in DES using convolutional neural networks
• TDCOSMO. XII. Improved Hubble constant measurement from lensing time delays using spatially resolved stellar kinematics of the lens galaxy
• Discovering New Strong Gravitational Lenses in the DESI Legacy Imaging Surveys
• An automatic taxonomy of galaxy morphology using unsupervised machine learning
• Galaxy Zoo: Morphological Classifications for 120,000 Galaxies in HST Legacy Imaging
• Deep learning for galaxy surface brightness profile fitting
• Deep Convolutional Neural Networks as strong gravitational lens detectors
• Galaxy morphological classification in deep-wide surveys via unsupervised machine learning
• Identifying Strong Lenses with Unsupervised Machine Learning using Convolutional Autoencoder
• Rates and Properties of Strongly Gravitationally Lensed Supernovae and their Host Galaxies in Time-Domain Imaging Surveys
• HOLISMOKES -- II. Identifying galaxy-scale strong gravitational lenses in Pan-STARRS using convolutional neural networks
• How to Find Gravitationally Lensed Type Ia Supernovae
• The BOSS Emission-Line Lens Survey. III. : Strong Lensing of Ly$\alpha$ Emitters by Individual Galaxies
• LensExtractor: A Convolutional Neural Network in Search of Strong Gravitational Lenses
• Survey of Gravitationally-lensed Objects in HSC Imaging (SuGOHI). II. Environments and Line-of-Sight Structure of Strong Gravitational Lens Galaxies to z ~ 0.8
• Survey of Gravitationally-lensed Objects in HSC Imaging (SuGOHI). VI. Crowdsourced lens finding with Space Warps
• HOLISMOKES -- I. Highly Optimised Lensing Investigations of Supernovae, Microlensing Objects, and Kinematics of Ellipticals and Spirals
• Photometric Redshift Estimation with a Convolutional Neural Network: NetZ
• Beyond the Hubble Sequence -- Exploring Galaxy Morphology with Unsupervised Machine Learning
• High-quality strong lens candidates in the final Kilo Degree survey footprint
• Strong lens systems search in the Dark Energy Survey using Convolutional Neural Networks
• Detecting outliers in astronomical images with deep generative networks
• HOLISMOKES. VI. New galaxy-scale strong lens candidates from the HSC-SSP imaging survey
• Mining for Strong Gravitational Lenses with Self-supervised Learning
• Survey of Gravitationally-lensed Objects in HSC Imaging (SuGOHI). V. Group-to-cluster scale lens search from the HSC-SSP Survey
• HOLISMOKES. VIII. High-redshift, strong-lens search in the Hyper Suprime-Cam Subaru Strategic Program
• Strong gravitational lensing and microlensing of supernovae
• Survey of Gravitationally-lensed Objects in HSC Imaging (SuGOHI). IV. Lensed quasar search in the HSC survey
• Fanaroff-Riley classification of radio galaxies using group-equivariant convolutional neural networks
• Deep Learning for Strong Lensing Search: Tests of the Convolutional Neural Networks and New Candidates from KiDS DR3
• HOLISMOKES -- IV. Efficient mass modeling of strong lenses through deep learning
• Anomaly detection in Hyper Suprime-Cam galaxy images with generative adversarial networks
• Quantifying Non-parametric Structure of High-redshift Galaxies with Deep Learning
• North Ecliptic Pole merging galaxy catalogue
• Prediction of Supernova Rates in Known Galaxy-galaxy Strong-lens Systems
• Strong lens modelling: comparing and combining Bayesian neural networks and parametric profile fitting
• Detecting gravitational lenses using machine learning: exploring interpretability and sensitivity to rare lensing configurations
• Spin Parity of Spiral Galaxies II: A catalogue of 80k spiral galaxies using big data from the Subaru Hyper Suprime-Cam Survey and deep learning
• GAlaxy Light profile convolutional neural NETworks (GaLNets). I. fast and accurate structural parameters for billion galaxy samples
• The impact of human expert visual inspection on the discovery of strong gravitational lenses
• 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
• A targeted search for strongly lensed supernovae with the Las Cumbres Observatory

Cited by in corpus (6)

• Euclid: The Early Release Observations Lens Search Experiment
• HOLISMOKES XIII: Strong-lens candidates at all mass scales and their environments from the Hyper-Suprime Cam and deep learning
• HOLISMOKES XV. Search for strong gravitational lenses combining ground-based and space-based imaging
• Accelerating lensed quasar discovery and modeling with physics-informed variational autoencoders
• Searching for strong lensing by late-type galaxies in UNIONS
• HOLISMOKES XVI: Lens search in HSC-PDR3 with a neural network committee and post-processing for false-positive removal