Research map / Image generation / Fast diffusion sampling
Multi-Objective Learning for Diffusion Models: A Statistical Theory under Semi-Supervised Learning
Ziheng Cheng, Yixiao Huang, Hanlin Zhu, Haoran Geng, Somayeh Sojoudi, Jitendra Malik, Pieter Abbeel, Xin Guo
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 specialist-generated pseudo-rollouts to generalist training | multi-task diffusion policy · StackCube Level 2 (OOD) | closed-loop success rate: 24% -> 40% | Improved |
| add specialist-generated pseudo-rollouts to generalist training | multi-task diffusion policy · StackCube Level 3 (OOD) | closed-loop success rate: 8% -> 22% | Improved |
| add specialist-generated pseudo-rollouts to generalist training | multi-task diffusion policy · PickCube Level 0 | closed-loop success rate: 25% -> 69% | Improved |
| add specialist-generated pseudo-rollouts to generalist training | multi-task diffusion policy · PickCube Level 3 (OOD) | closed-loop success rate: 26% -> 42% | Improved |
| add specialist-generated pseudo-rollouts to generalist training | multi-task diffusion policy · StackCube Level 0 and Level 1 | closed-loop success rate: 36% -> 43% (L0), 36% -> 44% (L1) | Improved |
| add specialist-generated pseudo-rollouts to generalist training | multi-task diffusion policy · PickCube Level 2 (OOD) | closed-loop success rate: 35% -> 41% | Improved |
| add specialist-generated pseudo-pairs to generalist training | RDDM inpainting generalist · CelebA-HQ inpainting, alternating lines | PSNR: 22.939 -> 24.351 | Improved |
| add specialist-generated pseudo-pairs to generalist training | RDDM inpainting generalist · CelebA-HQ inpainting, face mask | PSNR: 21.557 -> 22.340 | Improved |
| add specialist-generated pseudo-pairs to generalist training | RDDM inpainting generalist · CelebA-HQ inpainting, wide mask | PSNR: 20.988 -> 21.610 | Improved |
Closest papers by meaning
- Provable Sample-Efficient Transfer Learning Conditional Diffusion Models via Representation Learning (NeurIPS 2025 · similarity 0.87)
- Conditioning Matters: Training Diffusion Policies is Faster Than You Think (NeurIPS 2025 · similarity 0.87)
- Constrained Diffusion Models via Dual Training (NeurIPS 2024 · similarity 0.87)
- Gradient Guidance for Diffusion Models: An Optimization Perspective (NeurIPS 2024 · similarity 0.87)
- Prediction with Action: Visual Policy Learning via Joint Denoising Process (NeurIPS 2024 · similarity 0.87)
- Multi-Robot Motion Planning with Diffusion Models (ICLR 2025 · similarity 0.87)
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.
