Risk-Sensitive Reinforcement Learning with Function Approximation:
               A Debiasing Approach

                    Yingjie Fei 1 Zhuoran Yang 2 Zhaoran Wang 1

             Abstract                 risk-seeking objective and β < 0 induces a risk-averse one.
                                  It can also be seen that Vβ tends to the risk-neutral V as
   We study function approximation for episodic re-        β → 0. Risk-sensitive RL has been widely applied in be-
   inforcement learning w ...