Research map / Image generation / Autoregressive and latent generation
Generalization of Diffusion Models Arises with a Balanced Representation Space
Zekai Zhang, Xiao Li, Xiang Li, Lianghe Shi, Meng Wu, Molei Tao, Qing Qu
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 |
|---|---|---|---|
| detect memorization with prompt-free representation standard deviation instead of Ross et al. geometric method | memorization detection · CIFAR10-EDM | AUROC: 0.713 -> 0.998 | Improved |
| detect memorization with prompt-free representation standard deviation instead of Ross et al. geometric method | memorization detection · CIFAR10-EDM | TPR at 1% FPR: 0.013 -> 0.984 | Improved |
| detect memorization with representation standard deviation instead of Ross et al. | memorization detection · ImageNet-DiT | TPR at 1% FPR: 0.528 -> 0.912 | Improved |
| detect memorization with representation standard deviation instead of Ross et al. | memorization detection · LAION-SD1.4 | runtime (s): 0.545 -> 0.067 | Improved |
| use representation standard deviation instead of prompt-based Wen et al. detector | memorization detection · LAION-SD1.4 | AUROC: 0.986 -> 0.987 | No clear effect |
| use representation standard deviation instead of Hintersdorf et al. | memorization detection · LAION-SD1.4 | runtime (s): 0.009 -> 0.067 | Hurt |
| apply steering with a single layer instead of six layers (ablation) | representation steering on Stable Diffusion v1.4 · image editing | edit quality (qualitative) | No clear effect |
Closest papers by meaning
- On the Edge of Memorization in Diffusion Models (NeurIPS 2025 · similarity 0.93)
- Understanding diffusion models requires rethinking (again) generalization (NeurIPS 2026 · similarity 0.92)
- Learning Diffusion Models with Flexible Representation Guidance (NeurIPS 2025 · similarity 0.90)
- Why Diffusion Models Don’t Memorize: The Role of Implicit Dynamical Regularization in Training (NeurIPS 2025 · similarity 0.90)
- How Diffusion Models Memorize (NeurIPS 2026 · similarity 0.90)
- Does Generation Require Memorization? Creative Diffusion Models using Ambient Diffusion (ICML 2025 · similarity 0.90)
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.
