MCTS

Mastering the game of Go without human knowledge

What is the problem being addressed? The problem of developing a pure reinforcement learning approach (i.e. without the use of any domain knowledge beyond basic game rules) for the game of Go has been addressed.