Minimax
Evaluation Function e
e: State -> Number
how favorable is this state for MAX
e<0 -> good for MIN
e>0 -> good for MAX
Creating evaluation function
use weighted sum of features
features are extracted from state
Backing up values
compute MIN or MAX value for each node
MIN: choose smallest value of children
MAX: choose highest values of children
start at “leaves” -> then go up the tree
What does the backed up value mean for a node N?
it its the value of the best state MAX can reach from this node
Problem of this basic version?
At each turn we need to compute a deep tree. This can be improved by Pruning-
Idea Alpha-Beta Pruning
Try to figure out if we can ignore certain subtrees
We can discontiue below any MIN node where beta<=alpha
Algorithm Alpha-Beta Pruning
explore game tree to depth h using depth-first
update alpha/beta values of parent of node N if:
search under N has been finished
search was discontinued
prune branches
discontinue below MAX if: alpha(N) >= beta(any ancestor of N)
discontinue below MIN if: beta(N) <= alpha(any ancestor of N)
Benefits from Apha-Beta Pruning
best performance if children are ordered (decreasing/increasing backed up values depending on MAX/MIN node)
getting worse if children are ordered at random
use iterative deepending to order nodes
Zuletzt geändertvor 2 Jahren