Euclid: Searches for strong gravitational lenses using convolutional neural nets in Early Release Observations of the Perseus field ⋆

The EuclidWide Survey is predicted to find approximately 170 000 galaxy-galaxy strong lenses from its lifetime observation of 14 000 deg2 of the sky. Detecting this many lenses by visual inspection with professional astronomers and citizen scientists alone is infeasible. As a result, machine learning algorithms, particularly convolutional neural networks (CNNs), were developed and have proven fruitful in finding strong lenses, such that the usage of CNNs has increased. We identify the major challenge as the automatic detection of the smallest resolved galaxy-galaxy strong lenses while simultaneously maintaining a low false positive rate.We select all sources with VIS IE < 23 mag from the Euclid Early Release Observation imaging of the Perseus cluster. We apply a range of CNN architectures to detect strong lenses in these cutouts. One aim of this research is to have a quantified starting point on the achieved purity and completeness with our current version of CNN-based detection pipelines for the VIS images of the Euclid Wide Survey (EWS). All our networks perform extremely well on simulated data sets. However, when applied to real Euclid imaging, the highest lens purity is just ∼ 11%. Among all our networks, the false positives are typically identifiable by human volunteers as e.g. spiral galaxies, multiple sources, and artifacts, implying that improvements are still possible, perhaps via a second, more interpretable lens selection filtering stage. There is no alternative to human classification of CNN-selected lens candidates. Given the expected ∼ 105 lensing systems in Euclid, this implies ∼ 106 subjects for human classification, which while very large is not in principle intractable and not without precedent. Key words. Gravitational lensing: strong – Methods: statistical, machine learning – Methods: data analysis – Surveys