Uncertainty Estimation In Continuous Models Applied To Reinforcement Learning