Research map / Image generation / Fast diffusion sampling
Measurement-Consistent Langevin Corrector for Stabilizing Latent Diffusion Inverse Problem Solvers
Lee Hyoseok, Sohwi Lim, Eunju Cha, Tae-Hyun Oh
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add MCLC to the solver | LDPS · FFHQ Gaussian Deblur | FID: 100.10 -> 80.83 | Improved |
| add MCLC to the solver | PSLD · FFHQ Gaussian Deblur | FID: 89.18 -> 66.28 | Improved |
| add MCLC to the solver | ReSample · FFHQ Super Resolution 4x | FID: 70.16 -> 53.85 | Improved |
| add MCLC to the solver | LatentDAPS · FFHQ Gaussian Deblur | FID: 99.53 -> 100.58 | No clear effect |
| add MCLC instead of DiffStateGrad as plug-in | ReSample · FFHQ Super Resolution 4x | FID: 79.94 -> 53.85 (DiffStateGrad vs MCLC) | Improved |
| add DiffStateGrad (linear manifold assumption) as plug-in | ReSample · FFHQ Super Resolution 4x | FID: 70.16 -> 79.94 | Hurt |
| add MCLC under severe 92% random inpainting | ReSample · FFHQ 92% inpainting | PSNR: 21.51 -> 25.01 | Improved |
| add MCLC under heavier measurement noise (sigma 0.05) | ReSample · FFHQ 92% inpainting | PSNR: 21.26 -> 24.80 | Improved |
| add MCLC to a flow-based latent solver | FlowChef · FFHQ Motion Deblur | FID: 185.44 -> 100.40 | Improved |
| add MCLC to a recent latent solver | TReg · AFHQ-val 1K Gaussian Deblur | FID: 37.12 -> 27.62 | Improved |
| add measurement-consistent projection to Langevin corrector (ablation) | MCLC / LDPS · FFHQ 4x SR | y-PSNR: 31.59 -> 33.23 | Improved |
| increase the number of corrector steps (ablation) | MCLC / LDPS · FFHQ 92% random inpainting | PSNR, 1-LPIPS, y-PSNR | Mixed |
Closest papers by meaning
- Stability and Generalizability in SDE Diffusion Models with Measure-Preserving Dynamics (NeurIPS 2024 · similarity 0.89)
- Preconditioned Langevin Dynamics with Score-based Generative Models for Infinite-Dimensional Linear Bayesian Inverse Problems (NeurIPS 2025 · similarity 0.89)
- Solving Inverse Problems with Latent Diffusion Models via Hard Data Consistency (ICLR 2024 · similarity 0.87)
- SITCOM: Step-wise Triple-Consistent Diffusion Sampling For Inverse Problems (ICML 2025 · similarity 0.87)
- Inverse Problem Sampling in Latent Space Using Sequential Monte Carlo (ICML 2025 · similarity 0.87)
- Improving Diffusion Posterior Samplers with Lagged Temporal Corrections for Image Restoration (NeurIPS 2026 · similarity 0.87)
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.
