Research map / Learning theory and algorithms / Submodular and game algorithms
Payoff-Aware Prediction of Population Game Dynamics
Kang Wang, Xiao Wang, Bo Li, Renzhe Xu
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Studies payoff-aware prediction of population game dynamics, aiming to forecast population trajectories under a new payoff design before trajectory data for that design exist.
Problem
Predicting how aggregate population behavior evolves under a new payoff function when no trajectory data under that payoff are available; standard dynamics learning methods fit motion on observed trajectories and do not generalize across payoff changes.
Approach
Frames the task as payoff-aware prediction, where the learned model conditions on the payoff design so that trajectories can be predicted for unseen payoffs; the abstract is truncated so the specific mechanism is not stated.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| keep payoff explicit and learn payoff-independent cost structure (PA-RmD) instead of fitting realized motion with Neural ODE | dynamics predictor · zero-sum cyclic, held-out payoffs, K=3 | one-step error: (1.45 $\pm$ 0.05)E-2 -> (1.29 $\pm$ 0.05)E-3 | Improved |
| use PA-RmD instead of Neural ODE | dynamics predictor · potential congestion, held-out payoffs, K=3 | rollout RMSE: (2.07 $\pm$ 0.18)E-2 -> (5.83 $\pm$ 0.28)E-3 | Improved |
| use PA-RmD instead of environment-label methods (LEADS, CoDA) | dynamics predictor · zero-sum cyclic K=4 | rollout RMSE: (9.34 $\pm$ 0.99)E-2 -> (1.73 $\pm$ 0.09)E-2 | Improved |
| use PA-RmD instead of SINDy and SIAR | dynamics predictor · zero-sum cyclic K=3 | one-step error: (4.05 $\pm$ 0.00)E-2 -> (1.29 $\pm$ 0.05)E-3 | Improved |
| use PA-RmD instead of baselines | PA-RmD · YouTube Trending held-out countries | one-step and rollout error | Improved |
Closest papers by meaning
- Optimal Rates for Feasible Payoff Set Estimation in Games (ICML 2026 · similarity 0.87)
- Prediction-Aware Learning in Multi-Agent Systems (ICML 2025 · similarity 0.85)
- Evolutionary Prediction Games (NeurIPS 2025 · similarity 0.85)
- A-PSRO: A Unified Strategy Learning Method with Advantage Metric for Normal-form Games (ICML 2025 · similarity 0.85)
- Bounded Rationality Equilibrium Learning in Mean Field Games (AAAI 2025 · similarity 0.84)
- Convergence of No-Swap-Regret Dynamics in Self-Play (NeurIPS 2024 · similarity 0.84)
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.
