Final Year Project

Project Title

Using Deep CNN to estimate Photometric Redshift and explore feature space through visualization

Description

We will be developing a deep Convolutional Neural Network (CNN) that takes an input of five channeled UGRIZ image data that contains five bands representing different levels of information of a galaxy, and gives as output the re-shift of the galaxy. This is a step towards creating a map of the universe as we can position the galaxies in space using this network. Additionally, we will visualize the intermediate convolutional layers to understand the behavior of the network and explore the feature space of galaxy information.