Minimizing expected loss for risk-avoiding reinforcement learning | IEEE Conference Publication | IEEE Xplore