Diffusion-MRI is perfect for amortised inference: Same model needs to be inverted on every voxel, so an inference network has dramatic efficiency gains over MCMC, at same or better accuracy!
Fun collab @mackelab.bsky.social with @josepemanzano.bsky.social + @mristam.bsky.social!
New preprint led by @josepemanzano.bsky.social on amortised simulation-based inference #SBI to map orientation #uncertainty and perform brain #tractography. Great collaboration with @jakhmack.bsky.social, @ppjgoncalves.bsky.social, @deismic.bsky.social.
Diffusion-MRI is perfect for amortised inference: Same model needs to be inverted on every voxel, so an inference network has dramatic efficiency gains over MCMC, at same or better accuracy!
Fun collab @mackelab.bsky.social with @josepemanzano.bsky.social + @mristam.bsky.social!
New preprint led by @josepemanzano.bsky.social on amortised simulation-based inference #SBI to map orientation #uncertainty and perform brain #tractography. Great collaboration with @jakhmack.bsky.social, @ppjgoncalves.bsky.social, @deismic.bsky.social.
New preprint led by @josepemanzano.bsky.social on amortised simulation-based inference #SBI to map orientation #uncertainty and perform brain #tractography. Great collaboration with @jakhmack.bsky.social, @ppjgoncalves.bsky.social, @deismic.bsky.social.