: Finding strong gravitational lenses in the early release observations using convolutional neural networks

Several new galaxy-galaxy strong gravitational lenses have been detected in the early release observations (ERO) from Euclid. The all-sky survey is expected to find 170 000 new systems, which are expected to greatly enhancing studies of dark matter and dark energy, and to constrain the cosmological parameters better. As a first step, we visually inspect all galaxies in one of the ERO fields (Perseus) to identify candidate strong-lensing systems and compared them to the predictions from convolutional neural networks (CNNs). The entire ERO dataset is too large for an expert visual inspection, however. In this paper, we therefore extend the CNN analysis to the whole ERO dataset and use different CNN architectures and methods. Using five CNN architectures, we identified 8469 strong gravitational lens candidates from IE-band cutouts of 13 Euclid ERO fields and narrowed them down to 97 through visual inspection. The sample includes 14 grade A and 31 grade B candidates. We present the spectroscopic confirmation of a strong gravitational lensing candidate, EUCL J081705.61+702348.8. The foreground lensing galaxy, an early-type system at z = 0.335, and the background source, a star-forming galaxy at z = 1.475 with [O <ns0:sc>II</ns0:sc>] emission, are both identified. The lens modelling with the Euclid strong lens modelling pipeline revealed two distinct arcs in a lensing configuration, with an Einstein radius of 1$ \,.\!\!^{\prime \prime } $18 ± 0$ \,.\!\!^{\prime \prime } $03. This confirms the lensing nature of the system. These findings demonstrate that CNN-based candidate selection followed by visual inspection provides an effective approach for identifying strong lenses in Euclid data. They also highlight areas for improvement in future large-scale implementations.