Deep learning methods for obtaining photometric redshift estimations from images

Abstract

Knowing the redshift of galaxies is one of the first requirements of many cosmological experiments, and as it’s impossible to perform spectroscopy for every galaxy being observed, photometric redshift (photo-z) estimations are still of particular interest. Here, we investigate different deep learning methods for obtaining photo-z estimates directly from images, comparing these with ‘traditional’ machine learning algorithms which make use of magnitudes retrieved through photometry. As well as testing a convolutional neural network (CNN) and inception-module CNN, we introduce a novel mixed-input model which allows for both images and magnitude data to be used in the same model as a way of further improving the estimated redshifts. We also perform benchmarking as a way of demonstrating the performance and scalability of the different algorithms. The data used in the study comes entirely from the Sloan Digital Sky Survey (SDSS) from which 1 million galaxies were used, each having 5-filter (ugriz) images with complete photometry and a spectroscopic redshift which was taken as the ground truth. The mixed-input inception CNN achieved a mean squared error (MSE) =0.009, which was a significant improvement ($30{{\ \rm per\ cent}}$) over the traditional Random Forest (RF), and the model performed even better at lower redshifts achieving a MSE = 0.0007 (a $50{{\ \rm per\ cent}}$ improvement over the RF) in the range of z < 0.3. This method could be hugely beneficial to upcoming surveys such as Euclid and the Vera C. Rubin Observatory’s Legacy Survey of Space and Time (LSST) which will require vast numbers of photo-z estimates produced as quickly and accurately as possible.

The Dawes Review 10: The impact of deep learning for the analysis of galaxy surveys

“…In a number of subsequent works (Menou 2019;Henghes et al 2021;Zhou et al 2021), it was proposed to extend this pure convolutional approach to photometric redshift estimation to hybrid models, combining an MLP branch tasked with processed photometric features (e.g. colours) and a CNN branch having access to the full image of the galaxy.…”

Deep Machine Learning in Cosmology: Evolution or Revolution?

“…In our group we developed two decades ago an Artificial Neural Network software package starting from ANNz (Collister and Lahav 2004) and then ANNz2 (Sadeh et al 2016). We've also attempted to improve the photo-z by adding galaxy structural parameters (Soo et al 2018) and by a Deep Learning approach of analysing the full galaxy image (Henghes et al 2022, following Pasquet et al 2019. Furthermore, the photometric data can be used simultaneously to produce photo-z and stellar mass (Mucesh et al 2021).…”

“…Furthermore, the photometric data can be used simultaneously to produce photo-z and stellar mass (Mucesh et al 2021). We've also explored the scalability of ML algorithms for photo-z with the size of training data sets and other factors (Henghes et al 2022). For other ML approaches to the photo-z problem, e.g.…”

Machine Learning for Observational Cosmology

“…Refs. [103,104] show that CNN-based frameworks estimate photometric redshifts of SDSS galaxies with higher precision than traditional models that use integrated photometric information alone. Recently, an innovative image analysis model called Vision Transformer was proposed, which works as efficient and accurate as CNNs in estimating galaxy properties, especially when a large training dataset is available [105,106].…”