Research map / Image generation / Fast diffusion sampling
SPREAD: Sampling-based Pareto front Refinement via Efficient Adaptive Diffusion
Sedjro Salomon Hotegni, Sebastian Peitz
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 |
|---|---|---|---|
| use SPREAD diffusion refinement instead of gradient-based baselines | online MOO · RE21, RE33, RE34, RE41 | hypervolume: RE33 43.06 -> 133.76 vs PMGDA; RE41 936.17 -> 1008.75 vs HVGrad | Improved |
| use SPREAD on synthetic ZDT and DTLZ2 | online MOO · ZDT1-3, DTLZ2 | hypervolume: DTLZ2 22.97 (PMGDA) vs 22.91 | No clear effect |
| use SPREAD instead of baselines for diversity | online MOO · 11 online problems | Delta-spread: ZDT3 0.90 -> 0.53 | Improved |
| add repulsion and perturbation (diversity mechanisms) (ablation) | SPREAD · ZDT1/ZDT2/DTLZ7 | Delta-spread: +inf (w/o diversity) -> 0.32 on ZDT1 | Improved |
| add Gaussian RBF repulsion term (ablation) | SPREAD · ZDT1, DTLZ7, RE33 | hypervolume: ZDT1 4.25 -> 5.72; RE33 99.89 -> 133.76 | Improved |
| add repulsion on RE41 (ablation) | SPREAD · RE41 | hypervolume: 1011.03 (w/o repulsion) vs 1008.75 | No clear effect |
| add perturbation term (ablation) | SPREAD · ablation problems | hypervolume: RE21 69.01 -> 70.10; ZDT1 unchanged 5.72 | Mixed |
| keep only DDPM stochasticity (remove all diversity guidance) (ablation) | SPREAD · DTLZ2, RE21, RE41 | Delta-spread: DTLZ2 0.93 -> 0.73 | Mixed |
| use SPREAD for offline MOO with surrogate | Off-MOO-Bench · Synthetic and RE groups | average rank: Synthetic 3.50, RE 1.83 | Improved |
| use SPREAD instead of CDM-PSL in Bayesian MOO | MOBO · nine MOBO problems | log-hypervolume difference | Improved |
| use SPREAD rather than PMGDA in compute | online MOO · scalability (ZDT1, DTLZ2, RE41, DTLZ4) | computation time | Mixed |
Closest papers by meaning
- Preference-Guided Diffusion for Multi-Objective Offline Optimization (NeurIPS 2025 · similarity 0.91)
- ParetoFlow: Guided Flows in Multi-Objective Optimization (ICLR 2025 · similarity 0.89)
- Expensive Multi-Objective Bayesian Optimization Based on Diffusion Models (AAAI 2025 · similarity 0.87)
- Test-time Alignment of Diffusion Models without Reward Over-optimization (ICLR 2025 · similarity 0.87)
- DiRecT: Safe Diffusion-Based Planning via Receding-Horizon Denoising (NeurIPS 2026 · similarity 0.86)
- Amortized Active Generation of Pareto Sets (NeurIPS 2025 · similarity 0.86)
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.
