Self-generation of reward in reinforcement learning by universal rules of interaction with the external environment | IEEE Conference Publication | IEEE Xplore