Collected sources and patterns will appear here. Add from search, explore, or the patterns library.
(fMRISignal, Metadata) -> LatentRepresentation
Condition a diffusion transformer model on categorical experimental metadata (e.g., task type, resting state, disease status) during fMRI representation pretraining to unify heterogeneous datasets.
Problem it solves
Jointly training on diverse fMRI datasets with different brain states leads to misaligned latent spaces and representation drift.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.