Wow! Deepmind’s AlphaGO wins the third game as well to make it 3-0 against the forever world champ Lee Sedol! Great job by David Silver and team. What’s next Demis? From the Nature paper, it appears that some simple yet key features were explicitly programmed in along with the more general deep nets and reinforcement learning. Not clear about how much of domain knowledge went into the MCTS. Can they get rid of even that domain knowledge ?
How about a robot playing Golf with the top pros?
For more on game 3 of AlphaGO-lee Sedol, see https://gogameguru.com/alphago-shows-true-strength-3rd-victory-lee-sedol/