Plan Recognition (or what the hell are you trying to do?)
How does Messi know where to run when Iniesta holds the ball? How can we tell if a user is performing a fraud using our system? Even we, as human beings, don't always excel at these tasks. Plan Recognition is a key problem in AI that can be phrased as follows: When you observe someone else's actions (a user, a robot, a collaborator), what is that someone trying to do, and how? In this talk I will share my experience with this research problem, the unique challenges it can tackle and a use-case of my plan recognition work with a major financial company.