AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Metropolis Sampling for Constrained Diffusion Models

Nic Fishman, Leo Klarner, Emile Mathieu, Michael John Hutchinson, Valentin De Bortoli

NeurIPS 2023 · Image generation › Fast diffusion sampling · approach: Diffusion theory and design space · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use Metropolis forward process instead of reflected Brownian motionconstrained diffusion model · hypercube d=2test log-likelihood: 2.25 -> 2.32Improved
use Metropolis instead of reflected forward processconstrained diffusion model · hypercube d=3test log-likelihood: 3.77 -> 4.15Improved
use Metropolis instead of reflected forward processconstrained diffusion model · hypercube d=10test log-likelihood: 7.42 -> 10.80Improved
use Metropolis instead of reflected forward processconstrained diffusion model · simplex d=3test log-likelihood: 2.64 -> 3.23Improved
use Metropolis instead of reflected forward processconstrained diffusion model · simplex d=10test log-likelihood: 7.00 -> 7.81Improved
use Metropolis instead of reflected forward processconstrained diffusion model · hypercube d=10training time (hours): 10.1 -> 0.90Improved
increase dimensionality from 2 to 10Metropolis vs reflected gap · hypercubetest log-likelihood: 2.25 -> 2.32 at d=2; 7.42 -> 10.80 at d=10Improved
use Metropolis instead of reflected forward processconstrained diffusion model · Robotics (SPD x R^2)test log-likelihood: 8.39 -> 9.13Improved
use Metropolis instead of reflected forward processconstrained diffusion model · Robotics (SPD x R^2)training time (hours): 9.52 -> 1.36Improved
use Metropolis instead of reflected forward processconstrained diffusion model · Protein backbonestest log-likelihood: 15.20 -> 15.33Improved
use Metropolis instead of reflected forward processconstrained diffusion model · Protein backbonestraining time (hours): 24.80 -> 3.12Improved
add boundary constraint via Metropolis instead of unconstrained diffusionMetropolis diffusion · Wildfire data on S^2MMD: 0.1567 -> 0.1388Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy