Continuous Deep QLearning with Modelbased Acceleration

深度Q-Learning与model-based方法结合来解决连续动作问题 Model-freereinforcementlearninghasbeensuccessfullyappliedtoarangeofchallengingproblems,andhasrecentlybeenextendedtohandlelargeneuralnetworkpoliciesandvaluefunctions.However,thesamplecomplexityofmodelfreealgorithms,particularly