counterfactual regret

math

Counterfactual Regret Minimization

For information imperfect games, Counterfactual Regret Minization (CFR) is

an iterative algorithm that converges to a nash equilibrium in any finite two-player zero-sum game

• http://martin.zinkevich.org/publications/regretpoker.pdf
• http://mlanctot.info/files/papers/nips09mccfr.pdf
• usage with deep learning: https://arxiv.org/pdf/1811.00164.pdf
• with code: https://nn.labml.ai/cfr/index.html