Latent Mixture of Symmetries for Sample-Efficient Dynamic Learning

Learning dynamics is essential for model-based control and Reinforcement Learning in systems operating in changing environments, such as robotics, autonomous vehicles, and power systems.