State-Augmentation Transformations for Risk-Sensitive Reinforcement Learning

Shuai Ma, Jia Yuan Yu

In the framework of MDP, although the general reward function takes three arguments-current state, action, and successor state; it is often simplified to a function of two argument…