Section 5.

Uploaded on:
Category: Music / Dance
Heuristic assessment of the normal utility of a given amusement state ... Whole number worth for kind of board piece. Nature of amusement state given by aggregate of material ...
Slide 1

Section 5 Game Playing

Slide 2

Games as inquiry Board amusements (e.g., Chess, Checkers, Go) Two player diversions High spreading variables Opponent player Contingency issue Opponent is expected to give a valiant effort (our most exceedingly terrible) Game environment Accessible complete access to the earth state Deterministic Next state controlled by move and ebb and flow state Episodic : no; Static : yes (for us); Discrete : yes

Slide 3

Game pursuit issue Initial state Configuration of board at begin, and which player begins Operators Legal moves that a player can make Goal or "terminal" test objective states called terminal states Utility capacity quality worth for terminal states from guidelines of amusement not an assessment if diversion play achieves this state Mini-max calculation can be utilized to proliferate values up quest tree for evaluations of utility of non-terminal states

Slide 4

Game trees Root hub player 1 or "max" attempting to expand score Immediate offspring of root player 2, or "min" attempting to minimize score Next layer, max Next layer, min, and so forth. A layer or conceivable move by a player is an "utilize"

Slide 5

Problems Complexity too high for pragmatic amusement playing Could utilize Depth restricted pursuit Iterative developing Means that we can\'t by and large utilize minimax engendering of utility qualities Use heuristic assessment rather Quiescent States that are unrealistic to show extensive changes Next state can be a checkmate, not tranquil Horizon issue Failing to make a move on account of profundity point of confinement on hunt

Slide 6

Evaluation capacities Evaluation capacity: h(n) for recreations Heuristic appraisal of the normal utility of a given diversion state Material favorable position capacity in chess Integer esteem for sort of load up piece Quality of amusement state given by whole of material preferred standpoint estimations of all pieces Example of a weighted straight total EVAL(s) = w 1 p 1 + w 2 p 2 + w 3 p 3 + … w\'s are the weights; p\'s are the tallies of a piece sort E.g., p 1 = 3 pawns w 1 for pawn is 1; p 2 = 2 knight or rook, w 2 for knight or rook is 3

Slide 7

Pruning Eliminating branches from tree in inquiry Alpha-beta pruning with smaller than normal max don\'t take after branches that can\'t impact choice still precise inside assessment capacity Principle : "On the off chance that you have a thought that is unquestionably awful, don\'t require some investment to perceive how genuinely terrible it is." (P. H. Winston)

Slide 8

8 1 7 2 3 5 3 2 9 3 9 7 1 9 6 2 16 6 2 5 4 6 1 4 2 max min max Example from P. H. Winston

View more...