DeepNash
An incomplete-information game AI system developed by DeepMind, based on the R-NaD (Regularized Nash Dynamics) algorithm. It approximates Nash equilibrium through self-play reinforcement learning and has demonstrated mastery of the imperfect-information board game Stratego.
Timeline (last 90 days)
DeepNash was developed by the DeepMind research team to tackle the imperfect-information game Stratego.
DeepNash's core algorithm is based on the R-NaD (Regularized Nash Dynamics) method.
DeepNash does not rely on expert game records or human prior knowledge, and learns its strategy entirely from scratch through self-play.
All Facts (3)
DeepNash was developed by the DeepMind research team to tackle the imperfect-information game Stratego.
50%UnverifiedDeepNash's core algorithm is based on the R-NaD (Regularized Nash Dynamics) method.
50%UnverifiedDeepNash does not rely on expert game records or human prior knowledge, and learns its strategy entirely from scratch through self-play.
50%