160 likes | 301 Vues
This introduction to Artificial Intelligence covers advanced concepts in adversarial search, focusing on Alpha-Beta pruning techniques and their significance in optimizing game strategies. Key concepts include defining the parameters of Alpha (best choice for maximizing player) and Beta (best choice for minimizing player), as well as practical examples of pruning to avoid unproductive paths in game trees. Through understanding these methodologies, students can enhance their problem-solving skills in AI, particularly in game playing scenarios like Reversi.
E N D
Introduction to Artificial IntelligenceCS 438 Spring 2008 • Today • AIMA, Ch. 6 • More Adversarial Search • Next Tuesday • AIMA, Ch. 13 • Reasoning with Uncertainty
SOE Open House • Schedule • EB 2029 • Activity • Man vs Machine Reversi • Class folder • Open House
Assignment 3: Game Playing • It Tourney Time!
α-β pruning • Strategy • Any path that is being expanded that is clearly worse than any known path can be abandoned early.
α is the value of the best (i.e., highest-value) choice found so far at any choice point along the path for max If v is worse than α, max will avoid it prune that branch Define β similarly for min Why is it called α-β?
α-β pruning • α (alpha): Best choice for max • β (beta): Best choice for min if maxChoice >= β then prune if minChoice <= α then prune
α-β pruning example if maxChoice >= β then prune if minChoice <= α then prune
α-β pruning example if maxChoice >= β then prune if minChoice <= α then prune
α-β pruning example if maxChoice >= β then prune if minChoice <= α then prune
α-β pruning example if α >= β then prune if β <= α then prune
How many states can be avoided? • It depends on the order states are generated in • If the best moves for each player are generated first then significant pruning can occur • If the worst moves are generated first then NO pruning can occur