ACReL: Adversarial Conditional value-at-risk Reinforcement Learning. (arXiv:2109.09470v2 [cs.LG] UPDATED)

In the classical Reinforcement Learning (RL) setting, one aims to find a
policy that maximizes its expected return. This objective may be inappropriate
in safety-critical domains such as healthcare or autonomous driving, where
intrinsic uncertainties due to stochastic policies and environment variability
may lead to catastrophic failures. This can be addressed by using the
Conditional-Value-at-Risk (CVaR) objective to instill risk-aversion in learned
policies. In this paper, we propose Adversarial Cvar Reinforcement Learning
(ACReL), a novel adversarial meta-algorithm to optimize the …

arxiv learning reinforcement reinforcement learning risk value