1

Chapter 6 Adversarial Search

CS 461 Artificial Intelligence Pinar Duygulu Bilkent University, Spring 2008
Slides are mostly adapted from AIMA and MIT Open Courseware

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

Outline
Games Optimal decisions Minimax algorithm - pruning Imperfect, real-time decisions

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

Games
Multi agent environments : any given agent will need to consider the actions of other agents and how they affect its own welfare. The unpredictability of these other agents can introduce many possible contingencies There could be competitive or cooperative environments Competitive environments, in which the agents goals are in conflict require adversarial search these problems are called as games

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

Games
In game theory (economics), any multiagent environment (either cooperative or competitive) is a game provided that the impact of each agent on the other is significant AI games are a specialized kind - deterministic, turn taking, two-player, zero sum games of perfect information In our terminology deterministic, fully observable environments with two agents whose actions alternate and the utility values at the end of the game are always equal and opposite (+1 and 1)

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

Games history of chess playing

1949 Shannon paper originated the ideas 1951 Turing paper hand simulation 1958 Bernstein program 1955-1960 Simon-Newell program 1961 Soviet program 1966 1967 MacHack 6 defeated a good player 1970s NW chess 4.5 1980s Cray Bitz 1990s Belle, Hitech, Deep Thought, 1997 - Deep Blue - defeated Garry Kasparov
Spring 2008

CS461 Artificial Intelligence Pinar Duygulu

Game Tree search

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

Partial Game Tree for Tic-Tac-Toe

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

Game tree (2-player, deterministic, turns)

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

Optimal strategies
In a normal search problem, the optimal solution would be a sequence of moves leading to a goal state - a terminal state that is a win In a game, MIN has something to say about it and therefore MAX must find a contingent strategy, which specifies
MAXs move in the initial state, then MAXs moves in the states resulting from every possible response by MIN, then MAXs moves in the states resulting from every possible response by MIN to those moves

An optimal strategy leads to outcomes at least as good as any other strategy when one is playing an infallible opponent
CS461 Artificial Intelligence Pinar Duygulu Spring 2008

10

Minimax
Perfect play for deterministic games Idea: choose move to position with highest minimax value = best achievable payoff against best play E.g., 2-ply game:

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

11

Minimax value
Given a game tree, the optimal strategy can be determined by examining the minimax value of each node (MINIMAX-VALUE(n)) The minimax value of a node is the utility of being in the corresponding state, assuming that both players play optimally from there to the end of the game Given a choice, MAX prefer to move to a state of maximum value, whereas MIN prefers a state of minimum value

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

12

Minimax algorithm

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

13

Minimax
MINIMAX-VALUE(root) = max(min(3,12,8), min(2,4,6), min(14,5,2)) = max(3,2,2) =3

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

14

Properties of minimax
Complete? Yes (if tree is finite) Optimal? Yes (against an optimal opponent) Time complexity? O(bm) Space complexity? O(bm) (depth-first exploration) For chess, b 35, m 100 for "reasonable" games exact solution completely infeasible

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

15

Tree Player and Non-zero sum games

(+1 +2 +3)

(+1 +2 +3)

(-1 +5 +2)

(+1 +2 +3)

(+6 +1 +2)

(-1 +5 +2)

(+5 +4 +5)

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

16

- pruning
It is possible to compute the correct minimax decision without looking at every node in the game tree
MINIMAX-VALUE(root) = max(min(3,12,8), min(2,x,y), min(14,5,2)) = max(3,min(2,x,y),2) = max(3,z,2) =3 where z <=2

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

17

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

18

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

19

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

20

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

21

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

22

Properties of -
Pruning does not affect final result Good move ordering improves effectiveness of pruning With "perfect ordering," time complexity = O(bm/2)
doubles depth of search

A simple example of the value of reasoning about which computations are relevant (a form of metareasoning)

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

23

Why is it called -?
is the value of the best (i.e., highestvalue) choice found so far at any choice point along the path for max If v is worse than , max will avoid it
prune that branch

Define similarly for min

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

24

The - algorithm

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

25

The - algorithm

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

26

The - algorithm

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

27

- pruning

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

28

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

29

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

30

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

31

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

32

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

33

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

34

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

35

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

36

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

37

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

38

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

39

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

40

- pruning example

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

41

MAX

MIN

<=6

MAX

>=8

MINH
6

I
5

J
8
= agent

= opponent

42

MAX

>=6

MIN

<=2

MAX

>=8

MINH
6

I
5

J
8
= agent

L
2

M
1
= opponent

43

MAX

>=6

MIN

MAX

>=8

MINH
6

I
5

J
8
= agent

L
2

M
1
= opponent

Alpha-beta Pruning
MAX
6

44

MIN

beta cutoff

MAX

>=8

alpha cutoff

MINH
6

I
5

J
8
= agent

L
2

M
1
= opponent

45

Move generation

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

46

Resource limits
Suppose we have 100 secs, explore 104 nodes/sec 106 nodes per move Standard approach: cutoff test:
e.g., depth limit (perhaps add quiescence search)

evaluation function
= estimated desirability of position

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

47

Evaluation function

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

48

Min-Max

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

49

Evaluation functions
A typical evaluation function is a linear function in which some set of coefficients is used to weight a number of "features" of the board position. For chess, typically linear weighted sum of features Eval(s) = w1 f1(s) + w2 f2(s) + + wn fn(s) e.g., w1 = 9 with f1(s) = (number of white queens) (number of black queens), etc.

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

50

Evaluation function

"material", : some measure of which pieces one has on the board. A typical weighting for each type of chess piece is shown Other types of features try to encode something about the distribution of the pieces on the board.
CS461 Artificial Intelligence Pinar Duygulu Spring 2008

51

Cutting off search

MinimaxCutoff is identical to MinimaxValue except
1. 2. Terminal? is replaced by Cutoff? Utility is replaced by Eval

Does it work in practice? bm = 106, b=35 m=4 4-ply lookahead is a hopeless chess player!
4-ply human novice 8-ply typical PC, human master 12-ply Deep Blue, Kasparov

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

52

The key idea is that the more lookahead we can do, that is, the deeper in the tree we can look, the better our evaluation of a position will be, even with a simple evaluation function. In some sense, if we could look all the way to the end of the game, all we would need is an evaluation function that was 1 when we won and -1 when the opponent won.
CS461 Artificial Intelligence Pinar Duygulu Spring 2008

53

it seems to suggest that brute-force search is all that matters. And Deep Blue is brute indeed... It had 256 specialized chess processors coupled into a 32 node supercomputer. It examined around 30 billion moves per minute. The typical search depth was 13ply, but in some dynamic situations it could go as deep as 30.

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

54

Practical issues

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

55

Other games

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

56

Deterministic games in practice

Checkers: Chinook ended 40-year-reign of human world champion Marion Tinsley in 1994. Used a precomputed endgame database defining perfect play for all positions involving 8 or fewer pieces on the board, a total of 444 billion positions. Chess: Deep Blue defeated human world champion Garry Kasparov in a sixgame match in 1997. Deep Blue searches 200 million positions per second, uses very sophisticated evaluation, and undisclosed methods for extending some lines of search up to 40 ply. Othello: human champions refuse to compete against computers, who are too good. Go: human champions refuse to compete against computers, who are too bad. In go, b > 300, so most programs use pattern knowledge bases to suggest plausible moves.

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008

57

Summary
Games are fun to work on! They illustrate several important points about AI perfection is unattainable must approximate good idea to think about what to think about

CS461 Artificial Intelligence Pinar Duygulu

Spring 2008