The author wrote "The reinforcement learning (RL) is very close to human learning, because ... so RL is potentially one of the best approaches to creating [sic] really intelligent systems," But this kind of description should be more modest, like "very close to human learning" should be "close to human learning more or less," and "one of the best approaches" should be "somewhat of a good approach," for example. (Do you think the proposed agent is REALLY intelligent?)

Reference should be according to the order of appearance in this style, not alphabetical order of authors.

If author is to refer to a widely accepted theory or method, just cite the original. For example, Reinforcement Learning [11,13], in which [11] is by the author of this submission, would give readers an incorrect impression that the author is one of those originally  proposed the theory/method. Cite just [13] in this case.

The author fails to be clear  WHAT IS NEW IN THIS PAPER. The reviewer felt it as if reading a text book. 

The author wrote, "Results of experiments are described in details in [11],"  then what is this paper for?

English expression should be improved if this paper survives the review procedure.