AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Likelihood Matching for Diffusion Models

Lei Qian, Wu Su, Yanqi Huang, Song Xi Chen

ICML 2026 · Image generation › Fast diffusion sampling · approach: Diffusion theory and learned samplers

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 Likelihood Matching with diagonal Hessian (r=0) instead of score matchingdiffusion model on CIFAR-10 · CIFAR-10FID: 3.15 -> 3.12Improved
increase Hessian rank to r=20-30 (ablation)Likelihood Matching · CIFAR-10FID: 3.12 -> 3.01Improved
increase Hessian rank to r=100 or 200 (ablation)Likelihood Matching · CIFAR-10FID: 3.01 -> 3.09Hurt
use Likelihood Matching (r=30) instead of score matchingdiffusion model on CelebA · CelebA 64x64FID: 2.71 -> 2.62Improved
use Likelihood Matching (r=30) instead of score matchingdiffusion model on CIFAR-10 · CIFAR-10NLL (bpd): 3.28 -> 3.11Improved
learn the Hessian rather than fix it to identity (ablation)Likelihood Matching · MNIST (N=2)FID: 10.28 (identity Hessian) vs full model lowerImproved
increase number of transition steps N and Hessian rank r (ablation)Likelihood Matching · MNISTFIDImproved
use Likelihood Matching instead of score matching on Gaussian and t mixturesMLP score and Hessian models · 1D mixture (MMD)MMDImproved
use Likelihood Matching (N=2) instead of score matchingSwiss-roll distribution · 2D Swiss-roll (log-MMD)log-MMDImproved
use Likelihood Matching with Hessian-based sampler at few reverse stepsMNIST sampler · MNISTFIDImproved
use Likelihood Matching instead of score matching on ImageNet 224x224training cost · ImageNet 224x224training time and memory: 3-4x training time, 2-3x memoryHurt

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