Searching with predictions

predictions, we study search games in which the searcher enhances its
strategy with some (unreliable) hint about the hider's position in the
environment. The objective is to quantify the tradeoffs between the
performance of the strategy in settings in which the hint is trusted,
adversarially obtained or, more generally, has unknown, but bounded
error. I will discuss applications of this framework to line and star
search under pure strategies, but also some ongoing work on mixed
strategies for problems such as discrete and expanding search.