Research map / Probabilistic and scientific modeling / Variational and generative sampling
Manifold Matching via Deep Metric Learning for Generative Modeling
Mengyu Dai, Haibin Hang
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 |
|---|---|---|---|
| replace GAN loss with manifold matching (MvM) | ResNet generator · Set5 (x4 SISR) | PSNR: 29.03 -> 29.76 | Improved |
| replace GAN loss with manifold matching (MvM) | RDN generator · Set5 (x4 SISR) | PSNR: 29.07 -> 30.06 | Improved |
| replace GAN loss with manifold matching (MvM) | NSRNet generator · Set5 (x4 SISR) | PSNR: 29.46 -> 29.79 | Improved |
| replace GAN loss with manifold matching (MvM) | ResNet generator · Set5 (x4 SISR) | LPIPS: 0.1885 -> 0.1941 | Hurt |
| replace GAN loss with manifold matching (MvM) | RDN generator · Urban100 (x4 SISR) | LPIPS: 0.2876 -> 0.2566 | Improved |
| replace GAN loss with manifold matching (MvM) | NSRNet generator · Set5 (x4 SISR) | NIQE: 5.7818 -> 6.0846 | Hurt |
| use MvM instead of ESRGAN perceptual setup | SISR generator · Set5 (x4 SISR) | PSNR: 29.63 -> 29.87 | Improved |
| use MvM instead of ESRGAN perceptual setup | SISR generator · Set5 (x4 SISR) | LPIPS: 0.1227 -> 0.1281 | Mixed |
| use MvM instead of ESRGAN perceptual setup | SISR generator · Set5 (x4 SISR) | NIQE: 6.2838 -> 5.8904 | Improved |
| use MvM instead of WGAN-GP | ResNet generator · CelebA 64x64 | FID: 18.0 -> 11.1 | Improved |
| use MvM instead of SWGAN | ResNet generator · LSUN bedroom 64x64 | FID: 14.9 -> 13.7 | Improved |
| match both centroid and 2-diameter instead of either alone (ablation) | MvM · CelebA 32x32 | sample quality (qualitative) | Improved |
Closest papers by meaning
- What Secrets Do Your Manifolds Hold? Understanding the Local Geometry of Generative Models (ICLR 2025 · similarity 0.87)
- Distance Marching for Generative Modeling (NeurIPS 2026 · similarity 0.87)
- Detecting Generated Images by Fitting Natural Image Distributions (NeurIPS 2025 · similarity 0.85)
- Categorical Flow Matching on Statistical Manifolds (NeurIPS 2024 · similarity 0.85)
- Follow the Energy, Find the Path: Riemannian Metrics from Energy-Based Models (NeurIPS 2025 · similarity 0.85)
- Metric Flow Matching for Smooth Interpolations on the Data Manifold (NeurIPS 2024 · similarity 0.85)
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.
