Domain adaptation in application to gravitational lens finding

TL;DR

This paper evaluates three domain adaptation techniques to improve automated gravitational lens detection in real survey data, focusing on enhancing model transfer from simulated to observational data.

Contribution

It compares the effectiveness of ADDA, WDGRL, and SDA in adapting lens-finding models trained on simulations to real observations, highlighting the best approaches.

Findings

WDGRL with ENN encoder performs best in unsupervised adaptation.

Supervised domain adaptation improves false positive discrimination.

The methods enhance automated lens detection for upcoming surveys.

Abstract

The next decade is expected to see a tenfold increase in the number of strong gravitational lenses, driven by new wide-field imaging surveys. To discover these rare objects, efficient automated detection methods need to be developed. In this work, we assess the performance of three domain adaptation techniques -- Adversarial Discriminative Domain Adaptation (ADDA), Wasserstein Distance Guided Representation Learning (WDGRL), and Supervised Domain Adaptation (SDA) -- in enhancing lens-finding algorithms trained on simulated data when applied to observations from the Hyper Suprime-Cam Subaru Strategic Program. We find that WDGRL combined with an ENN-based encoder provides the best performance in an unsupervised setting and that supervised domain adaptation is able to enhance the model's ability to distinguish between lenses and common similar-looking false positives, such as spiral galaxies, which is crucial for future lens surveys.