Scaling Learning based Policy Optimization for Temporal Tasks via Dropout

Model-based approach for training feedback controllers in highly nonlinear environments using DT-STL and dropout to handle long-horizon temporal tasks efficiently.