Exploiting Partial Observability and Optimized Simple State Representations in Deep Q-Learning | Publicación