Testing Convolutional Neural Networks for finding strong gravitational lenses in KiDS

TL;DR

This paper develops and tests convolutional neural networks trained on KiDS survey data to efficiently identify strong gravitational lens candidates, achieving high accuracy and potential for large-scale future surveys.

Contribution

Introduces two ConvNet models trained on real and simulated KiDS data, demonstrating improved accuracy in gravitational lens detection over previous methods.

Findings

Single-band ConvNet achieves ~40% purity in lens candidate selection.

The models retrieve 75% of confirmed lenses in the sample.

Simulated source quality significantly impacts ConvNet performance.

Abstract

Convolutional Neural Networks (ConvNets) are one of the most promising methods for identifying strong gravitational lens candidates in survey data. We present two ConvNet lens-finders which we have trained with a dataset composed of real galaxies from the Kilo Degree Survey (KiDS) and simulated lensed sources. One ConvNet is trained with single \textit{r}-band galaxy images, hence basing the classification mostly on the morphology. While the other ConvNet is trained on \textit{g-r-i} composite images, relying mostly on colours and morphology. We have tested the ConvNet lens-finders on a sample of 21789 Luminous Red Galaxies (LRGs) selected from KiDS and we have analyzed and compared the results with our previous ConvNet lens-finder on the same sample. The new lens-finders achieve a higher accuracy and completeness in identifying gravitational lens candidates, especially the single-band ConvNet. Our analysis indicates that this is mainly due to improved simulations of the lensed sources. In particular, the single-band ConvNet can select a sample of lens candidates with $\sim40\%$ purity, retrieving 3 out of 4 of the confirmed gravitational lenses in the LRG sample. With this particular setup and limited human intervention, it will be possible to retrieve, in future surveys such as Euclid, a sample of lenses exceeding in size the total number of currently known gravitational lenses.