Bayesian inverse reinforcement learning

Inverse reinforcement learning (IRL) methods attempt to recover the reward function of an agent by observing its behavior. Given the large amount of uncertainty in the underlying reward function, it is often useful to model this function …