AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Inverting Data Transformations via Diffusion Sampling

Jinwoo Kim, Sékou-Oumar Kaba, Jiyun Park, Seunghoon Hong, Siamak Ravanbakhsh

ICML 2026 · Image generation › Fast diffusion sampling · approach: Diffusion theory and learned samplers · 0 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
add TIED test-time inversion with classifier-logit energyResNet18 on MNIST · affNIST test, Aff(2,R)accuracy (%): 55.48 -> 82.64Improved
use VAE ELBO energy instead of classifier-logit energy (ablation)TIED · affNIST test, Aff(2,R)accuracy (%): 82.64 -> 96.95Improved
add diffusion at high noise levels (TIED) instead of gradient-only Langevin or LieLAC sampling (ablation)Trivialized energy-based sampling · affNIST test, Aff(2,R), VAE energyaccuracy (%): LieLAC 94.38 -> TIED 96.95; Langevin 74.72 -> 96.95Improved
use TIED sampling vs trivialized kinetic Langevin on SO(10)Energy-based Lie group sampler · SO(10) synthetic energy -10 X_{1,1}^2sampling steps to convergence: 100 diffusion steps vs 100,000 Langevin stepsImproved
add TIED to pretrained DeepONet on transformed 1D heat equationDeepONet · 1D heat equation, SL(2,R) ⋉ H(1,R)relative L2 error: 0.690 -> 0.043Improved
add TIED to DeepONet on 1D Burgers equationDeepONet · 1D Burgers equation, SL(2,R) ⋉ (R2,+)relative L2 error: 0.867 -> 0.178Improved
add TIED to DeepONet trained with data augmentationDeepONet + data aug. · 1D heat equation, transformedrelative L2 error: 0.081 -> 0.052Improved
use FoCal Bayesian optimization on heat equationDeepONet + FoCal · 1D heat equation, SL(2,R) ⋉ H(1,R)relative L2 error: 0.690 -> 1.948Hurt
TIED with classifier energy versus specialized equivariant networksResNet18 + TIED · affNIST/homNISTaccuracy (%): VAE energy: 96.95 vs affConv 95.08; 97.43 vs homConv 95.71; classifier energy 82.64 and 89.84Mixed
use gradient-based energy methods with classifier energy under large affine transformationsKinetic Langevin, LieLAC · affNIST, classifier energyaccuracy (%): Langevin 53.98, LieLAC 73.79 vs ResNet18 55.48Hurt

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