Research map / Reinforcement learning / Visual and representation RL
Population-Free Pareto Tracking for Sample-Efficient Multi-Policy MORL
Zeyu Zhao, Yueling Che, Kaichen Liu, Jian Li, Junmei Yao
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 Stage 3 sparse-region filling (K=1 instead of K=0) (ablation) | MPFT-MOTD7 · Walker2d-2 | hypervolume (x10^7): 1.95 -> 2.04 | Improved |
| add Stage 3 sparse-region filling (K=1 instead of K=0) (ablation) | MPFT-MOTD7 · Walker2d-2 | sparsity (x10^4, lower better): 2.09 -> 1.13 | Improved |
| increase K from 1 to 6 (ablation) | MPFT-MOTD7 · Walker2d-2 | hypervolume (x10^7): 2.04 -> 2.15 -> 2.20 | Improved |
| increase K from 1 to 6 (ablation) | MPFT-MOTD7 · Walker2d-2 | sparsity (x10^4, lower better): 1.13 -> 1.09 -> 0.89 | Improved |
| add Stage 3 sparse-region filling (K=1 instead of K=0) (ablation) | MPFT-MOTD7 · HalfCheetah-2 | hypervolume (x10^6): 3.47 -> 3.51 | Improved |
| add Stage 3 sparse-region filling (K=1 instead of K=0) (ablation) | MPFT-MOTD7 · HalfCheetah-2 | sparsity (x10^3, lower better): 5.95 -> 2.38 | Improved |
| increase K from 1 to 6 (ablation) | MPFT-MOTD7 · HalfCheetah-2 | hypervolume (x10^6): 3.51 -> 3.52 -> 3.54 | Improved |
| increase K from 1 to 6 (ablation) | MPFT-MOTD7 · HalfCheetah-2 | sparsity (x10^3, lower better): 2.38 -> 1.10 -> 0.63 | Improved |
| use MPFT framework instead of evolutionary frameworks (MOTD7 variant) | MPFT-MOTD7 vs PGMORL/PA2D-MORL/C-MORL · <=3-objective MuJoCo envs | env_steps: decrease by 78.22% | Improved |
| use MPFT framework instead of evolutionary frameworks (MOSAC variant) | MPFT-MOSAC vs PGMORL/PA2D-MORL/C-MORL · <=3-objective MuJoCo envs | env_steps: decrease by 72.04% | Improved |
| use MPFT framework instead of evolutionary frameworks (MOPPO variant) | MPFT-MOPPO vs PGMORL/PA2D-MORL/C-MORL · <=3-objective MuJoCo envs | env_steps: decrease by 66.27% | Improved |
| replace C-MORL with single-policy Pareto tracking (SPFT) | SPFT · FT-6 | hypervolume (x10^4): 3.67 -> 5.02 | 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.
- PA2D-MORL — PA2D-MORL: Pareto Ascent Directional Decomposition Based Multi-Objective Reinforcement Learning (AAAI 2024) · better on 6 of 6 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
- Pareto Set Learning for Multi-Objective Reinforcement Learning (AAAI 2025 · similarity 0.90)
- PA2D-MORL: Pareto Ascent Directional Decomposition Based Multi-Objective Reinforcement Learning (AAAI 2024 · similarity 0.89)
- Preference Conditioned Multi-Objective Reinforcement Learning: Decomposed, Diversity-Driven Policy Optimization (NeurIPS 2026 · similarity 0.89)
- On Generalization Across Environments In Multi-Objective Reinforcement Learning (ICLR 2025 · similarity 0.87)
- Preference Controllable Reinforcement Learning with Advanced Multi-Objective Optimization (ICML 2025 · similarity 0.87)
- MODULI: Unlocking Preference Generalization via Diffusion Models for Offline Multi-Objective Reinforcement Learning (ICML 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.
