Measuring photometric redshifts using galaxy images and Deep Neural Networks

摘要

We propose a new method to estimate the photometric redshift of galaxies byusing the full galaxy image in each measured band. This method draws from thelatest techniques and advances in machine learning, in particular Deep NeuralNetworks. We pass the entire multi-band galaxy image into the machine learningarchitecture to obtain a redshift estimate that is competitive with the bestexisting standard machine learning techniques. The standard techniques estimateredshifts using post-processed features, such as magnitudes and colours, whichare extracted from the galaxy images and are deemed to be salient by the user.This new method removes the user from the photometric redshift estimationpipeline. However we do note that Deep Neural Networks require many orders ofmagnitude more computing resources than standard machine learningarchitectures.