Photometric redshifts from SDSS images using a ConvolutionalNeural Network

Johanna Pasquet(1), E. Bertin(2), M. Treyer(3), S. Arnouts(3)and D. Fouchez(1)

(1) Aix Marseille Univ, CNRS/IN2P3, CPPM, Marseille, France

(2) Sorbonne Université, CNRS, UMR 7095, Institut d’Astrophysique de Paris, 98 bis bd Arago, 75014 Paris, France

(3) Aix Marseille Univ, CNRS, CNES, LAM, Marseille, France

Abstract

We developed a Deep Convolutional Neural Network (CNN), used as a classifier, to estimate photometric redshifts and associatedprobability distribution functions (PDF) for galaxies in the Main Galaxy Sample of the Sloan Digital Sky Survey atz<0.4. Ourmethod exploits all the information present in the images without any feature extraction. The input data consist of 64×64 pixelugrizimages centered on the spectroscopic targets, plus the galactic reddening value on the line-of-sight. For training sets of 100k objectsor more (≥20% of the database), we reach a dispersionσMAD<0.01, significantly lower than the current best one obtained fromanother machine learning technique on the same sample. The bias is lower than 10−4, independent of photometric redshift. The PDFsare shown to have very good predictive power. We also find that the CNN redshifts are unbiased with respect to galaxy inclination,and thatσMADdecreases with the signal-to-noise ratio (SNR), achieving values below 0.007 for SNR>100, as in the deep stackedregion of Stripe 82. We argue that for most galaxies the precision is limited by the SNR of SDSS images rather than by the method.The success of this experiment at low redshift opens promising perspectives for upcoming surveys.

Learn more

Key words.keywords — methods: data analysis – techniques: image processing- galaxies: distance and redshifts – surveys