Why it matters

Expectimax handles stochastic games. Understanding shapes stochastic AI.

Advertisement

The architecture

Max nodes: pick best.

Chance nodes: expected value.

Leaf: evaluation.

Expectimax evaluationMax nodepick highestChance nodeE[value]Leafstatic evalAlpha-beta pruning doesn't directly apply; probabilistic bounds are trickier
Expectimax.
Advertisement

How it works end to end

Chance nodes: weighted by probability.

No alpha-beta directly.

Sample-based (MCTS) preferred at scale.