CNN Architecture Comparison for Radio Galaxy Classification

TL;DR

This paper compares various CNN architectures for radio galaxy classification, analyzing overfitting, performance metrics, and architectural factors, and proposes a ranking system for practical applicability.

Contribution

It provides a comprehensive comparison of CNN models for radio galaxy classification, addressing overfitting issues and evaluating practical performance metrics.

Findings

MCRGNet, Radio Galaxy Zoo, and ConvXpress balance accuracy and computational efficiency.

Receptive field, stride, and coverage significantly affect recognition performance.

Ensemble methods improve classification accuracy.

Abstract

The morphological classification of radio sources is important to gain a full understanding of galaxy evolution processes and their relation with local environmental properties. Furthermore, the complex nature of the problem, its appeal for citizen scientists and the large data rates generated by existing and upcoming radio telescopes combine to make the morphological classification of radio sources an ideal test case for the application of machine learning techniques. One approach that has shown great promise recently is Convolutional Neural Networks (CNNs). Literature, however, lacks two major things when it comes to CNNs and radio galaxy morphological classification. Firstly, a proper analysis of whether overfitting occurs when training CNNs to perform radio galaxy morphological classification using a small curated training set is needed. Secondly, a good comparative study regarding the practical applicability of the CNN architectures in literature is required. Both of these shortcomings are addressed in this paper. Multiple performance metrics are used for the latter comparative study, such as inference time, model complexity, computational complexity and mean per class accuracy. As part of this study we also investigate the effect that receptive field, stride length and coverage has on recognition performance. For the sake of completeness, we also investigate the recognition performance gains that we can obtain by employing classification ensembles. A ranking system based upon recognition and computational performance is proposed. MCRGNet, Radio Galaxy Zoo and ConvXpress (novel classifier) are the architectures that best balance computational requirements with recognition performance.