Tag
This paper introduces an unsupervised framework for modeling acquisition-related variability in structural connectomes using hybrid latent space modeling, eliminating the need for manual capacity tuning by architecturally annealing encoder outputs.