Research map / Bandits and online learning / Online convex optimization and games
Dynamic Discounted Counterfactual Regret Minimization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes dynamic discounting schemes for counterfactual regret minimization that adapt iteration weighting, rather than using fixed discount parameters as in CFR+ and DCFR, to speed convergence on imperfect-information games.
Problem
Counterfactual regret minimization (CFR) variants such as CFR+ and DCFR use fixed, predefined discounting of earlier iterations to speed convergence in imperfect-information games; the abstract indicates these fixed schemes are the limitation being addressed.
Approach
Replace fixed iteration-discount schedules with dynamic discounting that adapts the weighting of iterations during CFR, extending the non-uniform weighting idea behind CFR+ and DCFR.
Experimental setup
Metrics: convergence rate
Baselines: CFR CFR+ DCFR
Builds on: CFR+ DCFR
Closest papers by meaning
- Faster Game Solving via Hyperparameter Schedules (AAAI 2026 · similarity 0.93)
- Stable-Predictive Optimistic Counterfactual Regret Minimization (ICML 2019 · similarity 0.93)
- Deep (Predictive) Discounted Counterfactual Regret Minimization (AAAI 2026 · similarity 0.92)
- Solving Imperfect-Information Games via Discounted Regret Minimization (AAAI 2019 · similarity 0.92)
- AutoCFR: Learning to Design Counterfactual Regret Minimization Algorithms (AAAI 2022 · similarity 0.91)
- Accelerating Nash Equilibrium Convergence in Monte Carlo Settings Through Counterfactual Value Based Fictitious Play (NeurIPS 2024 · 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.