Speaker
Description
In the past, researchers have relied on single-resolution images from individual telescopes to detect gravitational lenses. We propose a search for galaxy-galaxy lenses that, for the first time, combines high-resolution single-band images (in our case Hubble Space Telescope, HST) with low-resolution multi-band images (in our case Legacy survey, LS) using machine learning. To compensate for the scarcity of lensed galaxy images for network training, we generated simulated lenses by superimposing arc features onto HST images, saved the lens parameters, and replicated the lens system in the LS images. We test four architectures based on ResNet-18: (1) using single-band HST images, (2) three-band LS images, (3) stacking these images after interpolation the LS images to HST resolution for simultaneous process, and (4) merging before the fully connected layer, a ResNet branch of HST with a ResNet branch of LS. Our results demonstrate that
models integrating images from both the HST and LS significantly enhance the detection of galaxy-galaxy lenses compared to models relying on data from a single instrument, and could be use in the future for LSST and Euclid.
| Abstract title | Search for strong gravitational lenses combining ground-based and space-based imaging |
|---|