Research map / Bandits and online learning / Online convex optimization and games
Stable-Predictive Optimistic Counterfactual 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 |
|---|---|---|---|
| Use OFTRL with the theoretically suggested stepsize instead of vanilla CFR | CFR · HUNL subgames 3 and 4, simultaneous updates | exploitability | Helped |
| Use OFTRL with the theoretically suggested stepsize instead of vanilla CFR | CFR · HUNL subgames 1 and 2, simultaneous updates | exploitability | Hurt |
| Tune the OFTRL stepsize by dividing it by 10, 100, or 1000 (ablation) | Stable-Predictive CFR · HUNL subgames 1 and 2, simultaneous updates | exploitability | Helped |
| Use OFTRL instead of vanilla CFR | CFR · HUNL subgames 1 and 2, alternating updates | exploitability | Mixed |
| Use manually tuned OFTRL instead of vanilla CFR (ablation) | CFR · HUNL subgames 3 and 4, alternating updates | exploitability | Hurt |
| Use alternating rather than simultaneous updates (ablation) | Stable-Predictive CFR · HUNL subgames 3 and 4 | exploitability | Hurt |
| Use DCFR instead of Stable-Predictive CFR | HUNL subgames · HUNL subgames 1–4, simultaneous updates | exploitability | Helped |
| Use DCFR instead of Stable-Predictive CFR | HUNL subgames · HUNL subgames 1–4, alternating updates | exploitability | Helped |
| Use stable-predictive regret minimizers rather than regular regret minimizers | CFR for two-player zero-sum EFGs · Two-player zero-sum EFGs | Nash equilibrium convergence rate: O(T^{-1/2}) -> O(T^{-3/4}) | Helped |
| Set each local regret minimizer's stability relative to its position in the decision tree and use counterfactual loss predictions | Stable-Predictive CFR · Sequential decision processes | regret bound | Helped |
Closest papers by meaning
- Dynamic Discounted Counterfactual Regret Minimization (ICLR 2024 · similarity 0.93)
- Online Convex Optimization for Sequential Decision Processes and Extensive-Form Games (AAAI 2019 · similarity 0.92)
- Faster Game Solving via Asymmetry of Step Sizes (AAAI 2026 · similarity 0.92)
- Solving Imperfect-Information Games via Discounted Regret Minimization (AAAI 2019 · similarity 0.92)
- Model-Free Online Learning in Unknown Sequential Decision Making Problems and Games (AAAI 2021 · similarity 0.92)
- Accelerating Nash Equilibrium Convergence in Monte Carlo Settings Through Counterfactual Value Based Fictitious Play (NeurIPS 2024 · similarity 0.90)
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.