On Imitation in Mean-field Games

Neural Information Processing Systems 

In this paper, departing from the existing literature on IL for MFGs, we introduce a new solution concept called the Nash imitation gap.