On machine learning search for gravitational lenses

TL;DR

This paper presents a machine learning approach using convolutional neural networks to detect gravitational lenses in sky images, achieving high accuracy on simulated data and successfully identifying a real Einstein cross as a lens.

Contribution

The study introduces a CNN-based method trained on realistic simulated images for gravitational lens detection, demonstrating its effectiveness on real sky data.

Findings

Achieved 93% classification accuracy on simulated images.

Successfully identified a real Einstein cross as a gravitational lens.

Proved the potential of machine learning for automated lens searches.

Abstract

We consider a machine learning algorithm to detect and identify strong gravitational lenses on sky images. First, we simulate different artificial but very close to reality images of galaxies, stars and strong lenses, using six different methods, i.e. two for each class. Then we deploy a convolutional neural network architecture to classify these simulated images. We show that after neural network training process one achieves about 93 percent accuracy. As a simple test for the efficiency of the convolutional neural network, we apply it on an real Einstein cross image. Deployed neural network classifies it as gravitational lens, thus opening a way for variety of lens search applications of the deployed machine learning scheme.