Research map / Image generation / Image restoration
Diffusion Representation for Asymmetric Kernels via Magnetic Transform
Mingzhen He, FAN He, Ruikai Yang, Xiaolin Huang
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 ME/MME with MagDM (uses negative spectrum) | MagDM · clustering, P=1 | adjusted Rand index: 0.2555±0.1818 -> 0.9266±0.0888 | Improved |
| Replace ME/MME with MagDM | MagDM · clustering, P=1 | normalized mutual information: 0.2124±0.1716 -> 0.9484±0.0667 | Improved |
| Replace ME with MagDM | MagDM · clustering, P=1 | Davies-Bouldin index: 0.4311±0.0848 -> 0.3924±0.0396 | Improved |
| Replace ME with MagDM | MagDM · clustering, P=1 | silhouette coefficient: 0.7987±0.0342 -> 0.8206±0.0423 | Improved |
| Replace ME with MagDM | MagDM · clustering, P=0 | adjusted Rand index: 0.9085±0.1051 -> 0.9365±0.0581 | Improved |
| Replace MME with MagDM | MagDM · clustering, P=0 | silhouette coefficient: 0.8616±0.0694 vs 0.8583±0.0461 | No clear effect |
| Replace MME with MagDM | MagDM · clustering, P=0 | Davies-Bouldin index: 0.3525±0.0587 vs 0.3647±0.0610 | No clear effect |
| Replace MME with MagDM | MagDM · clustering, P=0.2 | Davies-Bouldin index: 0.3012±0.0805 vs 0.3277±0.0717 | Mixed |
| Replace MME with MagDM | MagDM · clustering, P=0.2 | normalized mutual information: 0.8259±0.0871 -> 0.9106±0.1115 | Improved |
| Increase diffusion time t from 1 to 10 | MagDM · MagDM, four-group network | cluster separation (qualitative) | Hurt |
| Increase backward flow probability P toward 0.7-1 | MagDM · MagDM, eigenvalue signs | sign counts of top-2 eigenvalues: 100 -> 0 | Mixed |
| Use both positive and negative eigenvalues rather than truncating negatives | MagDM · MagDM vs ME/MME | group separation (qualitative) | Improved |
Closest papers by meaning
- Efficiently Access Diffusion Fisher: Within the Outer Product Span Space (ICML 2025 · similarity 0.84)
- Toward a Unified Geometry Understanding : Riemannian Diffusion Framework for Graph Generation and Prediction (NeurIPS 2025 · similarity 0.84)
- Efficient Diffusion Models for Symmetric Manifolds (ICML 2025 · similarity 0.83)
- Pard: Permutation-Invariant Autoregressive Diffusion for Graph Generation (NeurIPS 2024 · similarity 0.83)
- Equivariant spatio-hemispherical networks for diffusion MRI deconvolution (NeurIPS 2024 · similarity 0.83)
- SITCOM: Step-wise Triple-Consistent Diffusion Sampling For Inverse Problems (ICML 2025 · similarity 0.83)
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.
