A reinforcement learning algorithm based on model

The convergence rate is a very important aspect of machine learning.In reinforcement learning,the learning rate will be very low if the algorithms do not make full use of the experience knowledge in each process of reinforcement learning.In order to improve the convergence rate of reinforcement learning,a method in which the environment model learning process is introduced into reinforcement learning is presented.Firstly,the algorithm learns from the model of environment,and then it uses the obtained new model to guide the reinforcement learning process.Experiment results on the RoboCup emulation platform in the Linux system prove the validity of this algorithm.