AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

A Regularized Conditional GAN for Posterior Sampling in Image Recovery Problems

Matthew C Bendel, Rizwan Ahmad, Philip Schniter

NeurIPS 2023 · Image generation › Fast diffusion sampling · approach: Diffusion posterior sampling · 2 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 supervised-l1 plus SD reward regularization instead of Ohayon et al. lossconditional GAN · fastMRI R=4CFID (72 samples): 4.06 -> 3.10Improved
use proposed regularization instead of Ohayon et al. lossconditional GAN · fastMRI R=8CFID (72 samples): 6.04 -> 4.87Improved
use proposed regularization instead of Ohayon et al. lossconditional GAN · fastMRI R=4CFID (all 14576 samples): 2.95 -> 1.29Improved
use proposed cGAN instead of Adler & Oktem cGANconditional GAN · fastMRI R=8FID: 10.77 -> 7.72Improved
use proposed cGAN instead of Langevin samplingconditional GAN · fastMRI R=8FID: 14.32 -> 7.72Improved
use proposed cGAN instead of Langevin samplingconditional GAN · fastMRI4-sample generation time: 14 min -> 217 msImproved
compare APSD of Ohayon et al. cGAN with othersOhayon et al. cGAN · fastMRI R=4APSD: 7.2e-8 vs 3.8e-6Hurt
use proposed regularization (and remove MBSD) instead of CoModGAN (ablation)CoModGAN · CelebA-HQ inpaintingCFID: 5.29 -> 4.69Improved
use proposed regularization (and remove MBSD) instead of CoModGAN (ablation)CoModGAN · CelebA-HQ inpaintingFID: 8.50 -> 7.45Improved
use proposed cGAN instead of Score-SDEconditional GAN · CelebA-HQ inpaintingFID: 7.92 -> 7.45Improved
use proposed cGAN instead of Score-SDEconditional GAN · CelebA-HQ inpainting128-sample generation time: 48 min -> 217 msImproved
increase number of averaged samples P from 1 to 32proposed cGAN · fastMRI R=4PSNR: 36.96 -> 39.55Improved

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