3 Feb 2005
CS 3243 - Game playing
20
Evaluation functions
nFor
chess, typically linear weighted sum of features
nEval(s) = w1 f1(s) + w2 f2(s) + … + wn fn(s)
ne.g.,
w1 = 9 with
f1(s) = (number of white queens) – (number of black queens), etc.
nCaveat:
assumes independence of the features
nBishops
in chess better at endgame
nUnmoved
king and rook needed for castling
nShould model
the expected utility value states with the same feature values lead to.