Koopman-based surrogate modeling for reinforcement-learning-control of Rayleigh-Benard convection

Koopman-based surrogate modeling for reinforcement-learning-control of Rayleigh-Benard convection explores Accelerate reinforcement learning control of fluid dynamics systems by using surrogate models trained with policy-aware data, reducing training time by over 40% while maintaining state-of-the-art performance.. Commercial viability score: 4/10 in Reinforcement Learning Control.