Monte Carlo tree search

Terms from Artificial Intelligence: humans at the heart of algorithms

Page numbers are for draft copy at present; they will be replaced with correct numbers when final book is formatted. Chapter numbers are correct and will not change now.

Monte Carlo tree search is any form of Monte Carlo search applied to trees. For example a simple variant of depth search could randomly choose branches to follow, stopping when the path gets to a leaf, or, in the case of very large trees, at a random point before (getting bored!). As with any Monte Carlo techniques, heuristics can be used to assign probabilities to different branches. AlphaGo used Monte Carlo tree search alongside heuristics generated by deep neuaral networks.

Used in Chap. 7: page 106; Chap. 11: page 169