Research map / Image generation / Fast diffusion sampling
Efficient Diffusion Planning with Temporal Diffusion
Jiaming Guo, Rui Zhang, Zerun Li, Yunkai Gao, Shaohui Peng, Siming Lan, Xing Hu, Zidong Du, Xishan Zhang, Ling 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 |
|---|---|---|---|
| select top-m candidate plans instead of retaining all candidates (ablation) | TDP · Hopper (average over 3 datasets) | normalized score: 60.73 -> 93.7 | Improved |
| keep top-8 candidates instead of only the best plan (ablation) | TDP · Hopper (average over 3 datasets) | normalized score: 93.2 -> 93.7 | No clear effect |
| add automatic replanning instead of fixed replanning intervals (ablation) | TDP · Hopper | normalized score | Improved |
| use value-based criterion instead of state-based criterion for replanning (ablation) | TDP · Hopper | normalized score | Improved |
| add temporal diffusion (plan reuse) instead of Diffuser at long replanning intervals (ablation) | Diffuser · Hopper | normalized score | Improved |
| replan every step instead of at longer intervals (ablation) | TDP · Hopper Medium-Replay | normalized score | Hurt |
| use temporal diffusion plan refinement instead of replanning each step | Diffuser · Gym-MuJoCo and Kitchen | decision frequency: 1.4 -> 15.4 Hz | Improved |
| use temporal diffusion planner instead of Decision Diffuser | Decision Diffuser · Gym-MuJoCo and Kitchen | decision frequency: 0.62 -> 15.4 Hz | Improved |
| use TDP instead of Diffuser | Diffuser · D4RL locomotion average | normalized score: 77.5 -> 79.4 | Improved |
| use TDP instead of Diffuser | Diffuser · D4RL Kitchen average | normalized score: 53.1 -> 61.3 | Improved |
| use TDP instead of Decision Diffuser | Decision Diffuser · D4RL locomotion average | normalized score: 81.8 -> 79.4 | No clear effect |
| use inverse dynamics model (TDP_Inv) instead of TDP | TDP · D4RL locomotion average | normalized score: 79.4 -> 82.1 | Improved |
Closest papers by meaning
- Simple Hierarchical Planning with Diffusion (ICLR 2024 · similarity 0.89)
- DiRecT: Safe Diffusion-Based Planning via Receding-Horizon Denoising (NeurIPS 2026 · similarity 0.89)
- Prior-Guided Diffusion Planning for Offline Reinforcement Learning (NeurIPS 2025 · similarity 0.89)
- What Makes a Good Diffusion Planner for Decision Making? (ICLR 2025 · similarity 0.89)
- Habitizing Diffusion Planning for Efficient and Effective Decision Making (ICML 2025 · similarity 0.88)
- DiffuserLite: Towards Real-time Diffusion Planning (NeurIPS 2024 · 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.
