Formation and Evolution of
Galaxy Clusters Across Cosmic Time

In the recent years, the growth of astronomical datasets in volume and complexity has led to the diffusion of novel data analysis techniques, based on Machine and Deep Learning. These methods are particularly effective in extracting useful information and patterns from the datasets they are applied to, albeit not requiring any prior knowledge of them. In the near future, the need to further develop these algorithms will strongly increase because of the availability of the data that will be gathered in upcoming imaging surveys, such as that that will be carried out by the Euclid mission. Convolutional Neural Networks (CNNs), in particular, are a Deep Learning method that allows to efficiently process big volumes of images. In this work, we compare the performance of three Network architectures in the classification of galaxy-scale strong lenses on the basis of their morphological features, i.e. the distortion of the background source in wide arcs and rings around the lens galaxy. For this purpose, we apply our models to a dataset of 1e5 Euclid-like mock images simulated by the Bologna Lens Factory. We progressively consider larger fractions of faint lenses in the training sets and study the impact of this inclusion on the classification of the images. Our analysis confirms the potential of the application of this method for the identification of clear lenses, since our models find samples of these systems with > 90 % precision and completeness. On the other hand, we suggest that specific training for different classes of lenses might be needed for finding the faint lenses as well.