@NewYorker @nicolatwilley @anyonejanedoe love me some deepmind
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
.
@NewYorker@nicolatwilley I get referencing Deep Blue, but down the hall at IBM is Tesauro, who built TD-Gammon, the direct ancestor.#AI -
@mgershoff True, but I was deliberately contrasting chess, which people think of as intellectual, with Atari, which is seen as "mindless"... -
@nicolatwilley That's fair. Related, I think what is novel, and clearly not trivial, is the auto feature extraction, not the TD/Q-Learning. -
@mgershoff They're inseparable. It's the TD/Q-learning that makes the unsupervised feature extraction happen; DNN just makes it possible -
@nicolatwilley TD is a method to solve this equation - current reward plus discounted future stream of rewards.pic.twitter.com/FuYT60hice
-
@nicolatwilley My company,@conductrics, was built on top of RL - I built the prototype at a a machine learning summer school back in '08 -
@nicolatwilley The reason for the DNN, or CMACs, or RBF net, or whatever, is function approx, a model of the environment the agent lives in.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.