Top Banner
Alireza yousefpour [email protected]. ir
46

Introduction to game theory

Dec 03, 2015

Download

Documents

Ehsan Yavari

basics of game theory
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript

Slide 1

Alireza [email protected] Game Theory :

: : :

: . . ( .) : AI : . . . : - MIN-MAX . - (max) (min) .- max . max : : : (,) : ( ) : . : (1+) (1-) (0) : : Minimax :

: : : Minimax : : ( ) . :MAX MIN MAX MIN . MIN : Minimax :MIN-MAX . . .

MAX . MAX . 5 :27182718212718212271821This is the moveselected by min-maxStatic evaluator valueMaxMaxMinMaxMin2 : Minimax:: : minimax O(bm) . . : ( ) : O(bm)m: b: : : Minimax : : . . . : 35 50 35100 EVAL .- Coutoff test . - EVAL Minimax : : : . :Evaluation Function- . 0 Eval(n) 100 . ! ( minimax .) . . : :

: (horizon problem): . . : . .Eval(s) = w1 f1(s) + w2 f2(s) + + wnfn(s) ( ) :f w () . : d . . . : d : :

: : 1000 : = 35 150 150000 35 3 6 8 - : Alpha Beta Pruning : = 35 150 150000 35 3 6 8 - : Alpha Beta Pruning : : 1000 . ( ) minimax minimax . : : minimax . MAX . ( ) MIN . . h : -pruning

xij xi max(x1...xi-1) : xij+1

: -pruning xij xi min(x1...xi-1) : xij+1

: 453 Example:maxminmaxmin = 4 : 453170 Example:maxminmaxmin = 3 = 3 = 1 : 4531 86 Example:maxminmaxmin = 3 = 3 = 1 = 8-pruning70 = 3 : 4531 867 2641 Example:maxminmaxmin = 3 = 3 = 1 = 6 = 6 = 3 = 2 = 1 = 2 = 2 = 3 : : 4531 867 2641 Example:maxminmaxmin = 3 = 3 = 1 = 6 = 6 = 3 = 2 = 1 = 2 = 2 = 3

:[-, +][-,+] : 30

[-,3][-,+]

: : 31

[-,3][-,+]

: : 32

[3,+][3,3] : : 33

[-,2][3,+][3,3] Max : : 34

[-,2][3,14][3,3][-,14]

, : : 3536

[,2][3,5][3,3][-,5]

, : : 3637

[2,2][,2][3,3][3,3]

: : 3738

[2,2][-,2][3,3][3,3]

: : 38 - .1) O(b/log b) d .2) .3) backed-up .4) b b 5) minimax - : : : . A :

5-6 :(16-10 10-5) (24-19 11-5) (11-5 10-5) (16-11 11-5) : MIN MAX . . . . . : : A1 A2 .: min-max AMinDiceMax-12121013A1A2 : Expectimax (c) =i P(di) .maxs S(c,di) (utility(s)) S(C,di) P(di) C expectimax C : c .Expectimin (c) =i P(di) .mins S(c,di) (utility(s)) : AMinDiceMax-12121013A1A2 :Expectimin(A1) = 0.9* min{-1,2} + 0.1* min{1,2} = 0.9*(-1) + 0.1*(1) = -0.8Expectimin(A2) = 0.9* min{1,0} + 0.1* min{1,3} = 0.9*(0) + 0.1*(1) = 0.1A2 0.90.10.90.1 : expectiminimax O(bmnm) n . - . . 1+ 1- .: :