Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Model-Based Opponent Modeling
XiaoPeng Yu, Jiechuan Jiang, Wanpeng Zhang, Haobin Jiang, Zongqing Lu
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 MBOM instead of the strongest baseline, Meta-MAPG | agent policy · Triangle Game, fixed policy | score: -2.01 ± 0.06 -> -1.19 ± 0.03 | Improved |
| Use MBOM instead of the strongest baseline, Meta-MAPG | agent policy · Triangle Game, naïve learner | score: -5.76 ± 0.29 -> -1.66 ± 0.29 | Improved |
| Use MBOM instead of the strongest baseline, Meta-MAPG | agent policy · Triangle Game, reasoning learner | score: -6.14 ± 0.84 -> -2.75 ± 0.89 | Improved |
| Use MBOM instead of the strongest baseline, Meta-MAPG | agent policy · One-on-One, fixed policy | win rate (%): 44.4 ± 1.5 -> 59.5 ± 0.9 | Improved |
| Use MBOM instead of the strongest baseline, Meta-MAPG | agent policy · One-on-One, naïve learner | win rate (%): 36.6 ± 1.6 -> 51.3 ± 1.0 | Improved |
| Use MBOM instead of the strongest baseline, Meta-MAPG | agent policy · One-on-One, reasoning learner | win rate (%): 31.1 ± 1.9 -> 64.2 ± 1.8 | Improved |
| Add recursive imagination rather than using level-0 opponent modeling alone (ablation) | MBOM · Triangle Game and One-on-One, learning or reasoning opponents | performance | Improved |
| Add recursive imagination rather than using level-0 opponent modeling alone (ablation) | MBOM · Triangle Game and One-on-One, fixed-policy opponents | performance | No clear effect |
| Finetune IOPs with imagined best responses instead of random actions (ablation) | MBOM · Triangle Game and One-on-One, reasoning learners | performance | Improved |
| Add Bayesian mixing rather than use a single imagined policy (ablation) | MBOM · Triangle Game and One-on-One, reasoning learners | performance | Improved |
| Use Bayesian mixing instead of uniform mixing (ablation) | MBOM · Triangle Game and One-on-One | performance | Improved |
| Increase recursive-imagination levels from M=1 to M≥2 (ablation) | MBOM · Triangle Game and One-on-One | performance | Improved |
Closest papers by meaning
- An Open-Ended Learning Framework for Opponent Modeling (AAAI 2025 · similarity 0.89)
- Opponent Modeling based on Subgoal Inference (NeurIPS 2024 · similarity 0.89)
- Efficient Adaptation in Mixed-Motive Environments via Hierarchical Opponent Modeling and Planning (ICML 2024 · similarity 0.88)
- Offline Opponent Modeling with Truncated Q-driven Instant Policy Refinement (ICML 2025 · similarity 0.88)
- Generalized Intention Modeling in Multi-Agent Reinforcement Learning (NeurIPS 2026 · similarity 0.87)
- Opponent Modeling with In-context Search (NeurIPS 2024 · 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.
