Research map / Image generation / Fast diffusion sampling
ManifoldGD: Training-Free Hierarchical Manifold Guidance for Diffusion-Based Dataset Distillation
Ayush Roy, Wei-Yang Alex Lee, Rudrasis Chakraborty, Vishnu Suresh Lokhande
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 manifold guidance (tangent projection) to mode guidance | MGD · ImageNette IPC=10 ResNetAP-10 | accuracy (%): 61.9 -> 64.1 | Improved |
| add manifold guidance | MGD · ImageNette IPC=20 ResNetAP-10 | accuracy (%): 66.5 -> 69.7 | Improved |
| add manifold guidance | MGD · ImageNet-100 IPC=10 ResNetAP-10 | accuracy (%): 26.1 -> 27.6 | Improved |
| add manifold guidance | MGD · ImageWoof IPC=100 ResNet-18 | accuracy (%): 65.5 -> 66.3 | Improved |
| use divisive instead of agglomerative clustering for IPC selection (ablation) | ManifoldGD · ImageNette IPC=10 ConvNet-6 | accuracy (%): 37.7 -> 57.4 | Improved |
| use divisive instead of k-means for IPC selection (ablation) | ManifoldGD · ImageNette IPC=10 | accuracy (%): ConvNet-6 56.3 -> 57.4; ResNet-18 59.7 -> 58.5 | Mixed |
| add level-wise selection to divisive clustering (ablation) | ManifoldGD · ImageNette IPC=10 ResNet-18 | accuracy (%): 58.5 -> 61.1 | Improved |
| add manifold guidance on top of level-wise divisive (ablation) | ManifoldGD · ImageNette IPC=10 ResNet-18 | accuracy (%): 61.1 -> 62.3 | Improved |
| anneal neighborhood radius (ablation) | ManifoldGD · ImageNette IPC=10 ResNet-18 | accuracy (%): 62.3 -> 62.7 | Improved |
| increase guidance stop step TSTOP beyond about 25-30 of 50 (ablation) | ManifoldGD · ImageNette and ImageNet-100 | FID and accuracy | Hurt |
| add manifold guidance with Laplace kernel (ablation) | ManifoldGD · ImageNette IPC=10 ConvNet-6 | accuracy (%): 56.3 -> 57.0 | Improved |
| add manifold guidance with IMQ kernel (ablation) | ManifoldGD · ImageNette IPC=10 ResNet-18 | accuracy (%): 56.1 -> 56.7 | Improved |
Closest papers by meaning
- Manifold Preserving Guided Diffusion (ICLR 2024 · similarity 0.90)
- What Matters for Diffusion-Friendly Latent Manifold? Prior-Aligned Autoencoders for Latent Diffusion (NeurIPS 2026 · similarity 0.89)
- Unlocking Dataset Distillation with Diffusion Models (NeurIPS 2025 · similarity 0.89)
- CFG++: Manifold-constrained Classifier Free Guidance for Diffusion Models (ICLR 2025 · similarity 0.88)
- MGD$^3$ : Mode-Guided Dataset Distillation using Diffusion Models (ICML 2025 · similarity 0.88)
- Local Manifold Approximation and Projection for Manifold-Aware Diffusion Planning (ICML 2025 · similarity 0.88)
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.
