Home › Per AlphaGo (Silver et al. 2016, Nature), MONTE …
Per AlphaGo (Silver et al. 2016, Nature), MONTE CARLO TREE SEARCH (MCTS) explores a game tree using which strategy?
AUse exhaustive minimax with alpha-beta pruning
BRandom uniform sampling of moves without tree
CSelection + expansion + rollout + backpropagation
DAlways expand the deepest leaf to find a win
Answer & Solution
Correct answer: C. Selection + expansion + rollout + backpropagation
Silver et al. 2016 + Browne et al. 2012 MCTS survey: four phases per iteration. UCT (UCB1 applied to trees) balances exploitation + exploration over visits.
Related questions
In Reynolds-style steering behaviours combined into a STEERING PIPELINE, the OUTPUT accelePer Dave Mark + Brian Schwab, a RESPONSE CURVE in utility AI maps an input variable to a sIn modern game AI, REINFORCEMENT LEARNING agents (e.g. OpenAI Five, DeepMind AlphaStar) lePer Daniel Harabor's follow-up work and the public literature, JPS+ (the offline variant) Per Daniel Harabor (2011 AAAI paper), JPS (Jump Point Search) accelerates A* on UNIFORM-COIn tactical AI, an INFLUENCE MAP is best described as which data structure?In behaviour trees beyond sequence + selector, a DECORATOR node typically does which?Per Dave Mark ('Behavioral Mathematics for Game AI'), UTILITY AI selects actions by which