Research map / Image generation / Autoregressive and latent generation
FGM-HD: Boosting Generation Diversity of Fractal Generative Models through Hausdorff Dimension Induction
Haowei Zhang, Yuanpei Zhao, Ji-Zhe Zhou, Mao Li
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 HD loss with fixed weight (ablation) | FGM · ImageNet 256x256 | IS / Recall: IS 348.9 -> 333.7; Recall 0.46 -> 0.47 | Mixed |
| replace fixed HD loss weight with MMDS schedule (ablation) | FGM-HD · ImageNet 256x256 | Recall: 0.47 -> 0.51 | Improved |
| add HD rejection sampling alone (ablation) | FGM · ImageNet 256x256 | Recall: Recall 0.46 -> 0.58; FID 6.15 -> 6.78 | Mixed |
| combine MMDS with HD sampling (ablation) | FGM · ImageNet 256x256 | Recall: 0.46 -> 0.64 | Improved |
| MMDS vs fixed exponential schedule for HD loss weight (ablation) | FGM-HD training · training loss | final training loss: 3.14 -> 2.85 | Improved |
| raise HD sampling threshold (ablation) | FGM-HD · ImageNet 256x256 | FID / Recall | Mixed |
| use learned multi-scale ResNet152 HD estimator instead of box counting | HD estimation · Fractal dataset | time per image (s): 4.70 -> 0.32 | Improved |
| FGM-HD vs FGM overall | FGM · ImageNet 256x256 | Recall: 0.46 -> 0.64 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- DiffiT — DiffiT: Diffusion Vision Transformers for Image Generation (ECCV 2024) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Image Generation Diversity Issues and How to Tame Them (CVPR 2025 · similarity 0.85)
- Representation Fréchet Loss for Visual Generation (NeurIPS 2026 · similarity 0.85)
- HarmonyView: Harmonizing Consistency and Diversity in One-Image-to-3D (CVPR 2024 · similarity 0.83)
- ZoomLDM: Latent Diffusion Model for Multi-scale Image Generation (CVPR 2025 · similarity 0.83)
- ScaleCrafter: Tuning-free Higher-Resolution Visual Generation with Diffusion Models (ICLR 2024 · similarity 0.83)
- Improving Autoregressive Visual Generation with Cluster-Oriented Token Prediction (CVPR 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.
