1 paper
James McCarthy, Radu Marinescu, Elizabeth Daly +1
Risk-averse Constrained Reinforcement Learning (RaCRL) aims to learn policies that minimise the likelihood of rare and catastrophic constraint violations caused by an environment's…