Using a Neural Network Classifier to Select Galaxies with the Most Accurate Photometric Redshifts

Abstract

The Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST) will produce several billion photometric redshifts (photo-$z$'s), enabling cosmological analyses to select a subset of galaxies with the most accurate photo-$z$. We perform initial redshift fits on Subaru Strategic Program galaxies with deep $grizy$ photometry using Trees for Photo-Z (TPZ) before applying a custom neural network classifier (NNC) tuned to select galaxies with $(z_{\mathrm{phot}}-z_{\mathrm{spec}})/(1+z_{\mathrm{spec}})<0.10$. We consider four cases of training and test sets ranging from an idealized case to using data augmentation to increase the representation of dim galaxies in the training set. Selections made using the NNC yield significant further improvements in outlier fraction and photo-$z$ scatter (${\sigma }_z$) over those made with typical photo-$z$ uncertainties. As an example, when selecting the best third of the galaxy sample, the NNC achieves a 35% improvement in outlier rate and a 23% improvement in ${\sigma }_z$ compared to using uncertainties from TPZ. For cosmology and galaxy evolution studies, this method can be tuned to retain a particular sample size or to achieve a desired photo-$z$ accuracy; our results show that it is possible to retain more than a third of an LSST-like galaxy sample while reducing ${\sigma }_z$ by a factor of two compared to the full sample, with one-fifth as many photo-$z$ outliers. For surveys like LSST that are not limited by shot noise, this method enables a larger number of tomographic redshift bins and hence a significant increase in the total signal-to-noise of galaxy angular power spectra.

arXiv comment17 pages, 13 figures; Accepted to ApJ August 2021

References in corpus (43)

• Adam: A Method for Stochastic Optimization
• On the difficulty of training Recurrent Neural Networks
• LSST: from Science Drivers to Reference Design and Anticipated Data Products
• The Eleventh and Twelfth Data Releases of the Sloan Digital Sky Survey: Final Data from SDSS-III
• Halo Models of Large Scale Structure
• EAZY: A Fast, Public Photometric Redshift Code
• Accurate photometric redshifts for the CFHT Legacy Survey calibrated using the VIMOS VLT Deep Survey
• The COSMOS2015 Catalog: Exploring the 1<z<6 Universe with half a million galaxies
• The Hyper Suprime-Cam SSP Survey: Overview and Survey Design
• 3D-HST WFC3-selected Photometric Catalogs in the Five CANDELS/3D-HST Fields: Photometry, Photometric Redshifts and Stellar Masses
• Measuring and Modelling the Redshift Evolution of Clustering: the Hubble Deep Field North
• LSST Science Book, Version 2.0
• The Luminosity and Color Dependence of the Galaxy Correlation Function
• Large-Scale Galaxy Bias
• The DEEP2 Galaxy Redshift Survey: Design, Observations, Data Reduction, and Redshifts
• The 3D-HST Survey: Hubble Space Telescope WFC3/G141 grism spectra, redshifts, and emission line measurements for $\sim 100,000$ galaxies
• Galaxy And Mass Assembly (GAMA): end of survey report and data release 2
• The WiggleZ Dark Energy Survey: Survey Design and First Data Release
• Second Data Release of the Hyper Suprime-Cam Subaru Strategic Program
• The Sloan Digital Sky Survey Quasar Catalog: Fourteenth Data Release
• The VIMOS VLT Deep Survey final data release: a spectroscopic sample of 35016 galaxies and AGN out to z~6.7 selected with 17.5<=i_{AB}<=24.7
• Stable Architectures for Deep Neural Networks
• Photometric Redshifts for Hyper Suprime-Cam Subaru Strategic Program Data Release 1
• The PRIsm MUlti-Object Survey (PRIMUS) I: Survey Overview and Characteristics
• Core Cosmology Library: Precision Cosmological Predictions for LSST
• Measuring the Redshift Evolution of Clustering: the Hubble Deep Field South
• TPZ : Photometric redshift PDFs and ancillary information by using prediction trees and random forests
• PHAT: PHoto-z Accuracy Testing
• The VIMOS Public Extragalactic Survey (VIPERS): First Data Release of 57 204 spectroscopic measurements
• The sizes, masses and specific star-formation rates of massive galaxies at 1.3<z<1.5: strong evidence in favour of evolution via minor mergers
• The Impact of Non-Gaussian Errors on Weak Lensing Surveys
• The PRIsm MUlti-object Survey (PRIMUS). II. Data Reduction and Redshift Fitting
• The FMOS-COSMOS survey of star-forming galaxies at z~1.6 III. Survey design, performance, and sample characteristics
• High velocity outflows from young star-forming galaxies in the UKIDSS Ultra-Deep Survey
• The Arizona CDFS Environment Survey (ACES): A Magellan/IMACS Spectroscopic Survey of the Chandra Deep Field South
• The DEEP3 Galaxy Redshift Survey: Keck/DEIMOS Spectroscopy in the GOODS-N Field
• The FMOS-COSMOS survey of star-forming galaxies at $z\sim 1.6$ VI: Redshift and emission-line catalog and basic properties of star-forming galaxies
• Tomographic galaxy clustering with the Subaru Hyper Suprime-Cam first year public data release
• Data augmentation for machine learning redshifts applied to SDSS galaxies
• FMOS near-IR spectroscopy of Herschel selected galaxies: star formation rates, metallicity and dust attenuation at z~1
• Stellar polytropes and Navarro-Frenk-White halo models: comparison with observations
• Angular Correlation Function Estimators Accounting for Contamination from Probabilistic Distance Measurements
• Modeling the Gaia Color-Magnitude Diagram with Bayesian Neural Flows to Constrain Distance Estimates