One of the most important problems in machine learning—and life—is the exploration-exploitation dilemma. If you’ve found something that works, should you just keep doing it? Or is it better to try new things, knowing it could be a waste of time but also might lead to a better solution?
