There's a great Google Tech Talk by Geoff Hinton that serves as a slightly technical introduction to Deep Learning if anyone's interested:
https://www.youtube.com/watch?v=VdIURAu1-aU
ok; so I read the DeepMind paper and am totally not-impressed. But perhaps I'm confused. Does experience replay not a priori require a known system function? And if you already have a known system function what is the value of an algorithm to deduce what you already know?