Research map / Learning theory and algorithms / Combinatorial and search optimization
Pareto-Conditioned Diffusion Models for Offline Multi-Objective Optimization
Jatan Shrestha, Santeri Heiskanen, Kari Hepola, Severi Rissanen, Pekka Jääskeläinen, Joni Pajarinen
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 multi-objective reweighting to Ref. Dir. conditioning (ablation) | PCD · MO-Swimmer-v2, RE34, C10/MOP2 | HV: 3.53 -> 3.69 (Swimmer); 10.11 -> 10.17 (RE34); 10.47 -> 10.59 (C10/MOP2) | Improved |
| add reweighting to Ref. Dir. conditioning (ablation) | PCD · ZDT2 and Regex | HV: 7.89 -> 6.25 (ZDT2); 5.55 -> 4.80 (Regex) | Hurt |
| use reweighting instead of pruning low-quality points (ablation) | PCD · ZDT2 | HV: 5.64 -> 6.25 | Improved |
| use reweighting instead of pruning low-quality points (ablation) | PCD · RE34 and C10/MOP2 | HV: 10.16 -> 10.17 (RE34); 10.55 -> 10.59 (C10/MOP2) | Improved |
| use reference-direction conditioning instead of Ideal (ablation) | PCD without reweighting · MO-Swimmer-v2 | HV: 1.76 -> 3.53 | Improved |
| use reference-direction conditioning instead of D(best) points (ablation) | PCD without reweighting · RE34 | HV: 9.32 -> 10.11 | Improved |
| raise classifier-free guidance scale above 2.5 (ablation) | PCD · five representative tasks | normalized HV | No clear effect |
| increase reweighting temperature tau above 0.05 (ablation) | PCD · ZDT2 | normalized HV: up to 15% | Improved |
| increase reweighting temperature tau (ablation) | PCD · C10/MOP2 | normalized HV | No clear effect |
| use stochastic EDM sampler instead of deterministic (ablation) | PCD · MO-Hopper-v2, RE34, ZDT2 | HV: 4.47 -> 5.95 (Hopper); 9.47 -> 10.17 (RE34); 5.00 -> 6.25 (ZDT2) | Improved |
| use stochastic EDM sampler instead of deterministic (ablation) | PCD · Regex | HV: 5.56 -> 4.80 | Hurt |
| increase number of selected conditioning points J from 4 to 32 (ablation) | PCD · RE34 | HV: 6.56 -> 10.17 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ParetoFlow — ParetoFlow: Guided Flows in Multi-Objective Optimization (ICLR 2025) · better on 3 of 4 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
- Preference-Guided Diffusion for Multi-Objective Offline Optimization (NeurIPS 2025 · similarity 0.92)
- ParetoFlow: Guided Flows in Multi-Objective Optimization (ICLR 2025 · similarity 0.91)
- Offline Multi-Objective Optimization (ICML 2024 · similarity 0.90)
- Pareto Front-Diverse Batch Multi-Objective Bayesian Optimization (AAAI 2024 · similarity 0.88)
- PSMGD: Periodic Stochastic Multi-Gradient Descent for Fast Multi-Objective Optimization (AAAI 2025 · similarity 0.88)
- Expensive Multi-Objective Bayesian Optimization Based on Diffusion Models (AAAI 2025 · 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.
