EPOpt: Learning Robust Neural Network Policies Using Model Ensembles
Sample complexity and safety are major challenges when learning policies with reinforcement learning for real-world tasks, especially when the policies are represented using rich function approximators like deep neural …