PRISM: Diversifying Dataset Distillation by Decoupling Architectural Priors
TMLR
We introduce PRISM, a framework that disentangles architectural priors for dataset distillation, outperforming single-teacher setups.
A Study in Dataset Distillation for Image Super-Resolution
Accepted to ICPR 2026
We conduct the first systematic study of dataset distillation for Super-Resolution.
tobias-dietz