This evaluation compares convolutional and variational autoencoders for galaxy deblending, suggesting potential improvements.
Galaxy deblending is a critical challenge in astronomical surveys, particularly in large-scale surveys where overlapping galaxies complicate accurate observation and analysis. This paper evaluates autoencoder-based deep learning models for galaxy deblending, comparing convolutional autoencoders (CAE), Variational Autoencoders (VAE), and VAE-Generative Adversarial Networks. We find that CAEs have the best performance. We discuss the performance of the models as well as possible areas of improvements.
No takes yet. Share an insight, caveat, or question.
Narang et al. (2025) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: