Machines Learn Better if We Teach Them the Basics

A wave of research improves reinforcement learning algorithms by pre-training them as if they were human. Read more at QuantaMagazine.org. Music is “Quasi Motion” by Kevin MacLeod.