Research map / Bandits and online learning / Online convex optimization and games
Equilibrium Finding in Normal-Form Games via Greedy Regret Minimization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Choose each iteration's weight at runtime to minimize the regret potential | Regret minimization · Sampled two-player zero-sum random games | exploitability: almost an order of magnitude | Helped |
| Choose each iteration's weight at runtime to minimize the regret potential | Internal regret minimization · Sampled seven-player general-sum random games | internal regret: several orders of magnitude faster | Helped |
| Choose each iteration's weight at runtime to minimize the regret potential | Regret minimization · Random general-sum games | correlated-equilibrium convergence: dramatically improves convergence | Helped |
| Choose each iteration's weight at runtime to minimize the regret potential | Regret minimization · Diplomacy first-turn subgames | coarse-correlated-equilibrium convergence: several orders of magnitude | Helped |
| Choose each iteration's weight at runtime to minimize the regret potential | Regret minimization · Diplomacy first-turn subgames | correlated-equilibrium convergence | Helped |
| Choose each iteration's weight at runtime to minimize the regret potential | Regret minimization · France vs. Austria | Nash-equilibrium convergence: 2-3x fewer iterations | Helped |
| Choose each iteration's weight at runtime to minimize the regret potential | Regret minimization · France vs. Austria | Nash-equilibrium convergence: 2-3x faster | Helped |
| Use greedy weights instead of vanilla weighting | Regret matching · Random 7-player, 10-action general-sum games | average social welfare: 3.50 -> 4.16 | Helped |
| Use a relatively low weight floor rather than a high floor (ablation) | Greedy weights in general-sum games · Random general-sum games | regret convergence | Helped |
| Add a weight floor of 50% to 100% of the average iteration weight (ablation) | Greedy weights with external regret minimization · Two-player zero-sum games | minimax-equilibrium convergence: 50% to 100% | Helped |
| Use a low weight floor rather than a high floor (ablation) | Greedy weights with internal regret minimization · Two-player zero-sum games | regret convergence | Helped |
| Consider a small set of candidate weights instead of computing the exact optimal weight (ablation) | Greedy weights · Random seven-player general-sum games | regret convergence: several orders of magnitude | Helped |
Closest papers by meaning
- Regret Minimization in Behaviorally-Constrained Zero-Sum Games (ICML 2017 · similarity 0.90)
- Optimism Without Regularization: Constant Regret in Zero-Sum Games (NeurIPS 2025 · similarity 0.88)
- Mediator Interpretation and Faster Learning Algorithms for Linear Correlated Equilibria in General Sequential Games (ICLR 2024 · similarity 0.88)
- Policy-Regret Minimization in Markov Games with Function Approximation (ICML 2025 · similarity 0.88)
- Learning Not to Regret (AAAI 2024 · similarity 0.88)
- No Coin Left Behind: Maximizing Strategic Surplus Against No-Regret Dynamics (NeurIPS 2026 · 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.