Photometric redshift estimation via deep learning

Abstract

Context. The need to analyze the available large synoptic multi-band surveys drives the development of new data-analysis methods. Photometric redshift estimation is one field of application where such new methods improved the results, substantially. Up to now, the vast majority of applied redshift estimation methods have utilized photometric features. Aims. We aim to develop a method to derive probabilistic photometric redshift directly from multi-band imaging data, rendering pre-classification of objects and feature extraction obsolete. Methods. A modified version of a deep convolutional network was combined with a mixture density network. The estimates are expressed as Gaussian mixture models representing the probability density functions (PDFs) in the redshift space. In addition to the traditional scores, the continuous ranked probability score (CRPS) and the probability integral transform (PIT) were applied as performance criteria. We have adopted a feature based random forest and a plain mixture density network to compare performances on experiments with data from SDSS (DR9). Results. We show that the proposed method is able to predict redshift PDFs independently from the type of source, for example galaxies, quasars or stars. Thereby the prediction performance is better than both presented reference methods and is comparable to results from the literature. Conclusions. The presented method is extremely general and allows us to solve of any kind of probabilistic regression problems based on imaging data, for example estimating metallicity or star formation rate of galaxies. This kind of methodology is tremendously important for the next generation of surveys.

The many flavours of photometric redshifts

“…Finally, Deep Machine Learning (DML), the current state of the art in computer science, has been also applied to problem of photo-z estimation in [e.g. 51,52]. DML is based on normal neural networks, but with many thousands of neurons in each hidden layer.…”

Photometric redshifts from SDSS images using a convolutional neural network

“…aanda 2018), classification of light curves (Charnock & Moss 2017;Pasquet-Itam & Pasquet 2017). Thanks to the speed boost from Graphic Processing Units (GPU) technology and large galaxy spectroscopic sample such as the SDSS survey, Hoyle (2016);D'Isanto & Polsterer (2018) showed that Deep Convolutional neural network (CNN) were able to provide accurate phototometric redshifts from multichannel images, instead of extracted features, taking advantage of all the information contained in the pixels, such as galaxy surface brightness and size, disk inclination, or the presence of color gradients and neighbors. To do so, Hoyle (2016) used a Deep CNN inspired by the architecture of Krizhevsky et al (2012a), on 60×60 RGBA images, encoding colors (i − z, r − i, g − r) in RGB layers and r band magnitudes in the alpha layer.…”

XMMPZCAT: A catalogue of photometric redshifts for X-ray sources

“…SDSS, Beck et al 2016) and optical QSOs (e.g. Ball et al 2007Ball et al , 2008D'Isanto & Polsterer 2018). For X-ray AGN, though, only SED fitting techniques had been used (Salvato et al 2009;Hsu et al 2014), up until recently.…”