Uncertain Reasoning in Games Dmitrijs Rutko Faculty of Computing University of Latvia LU and LMT Computer Science Days at Ratnieki, 2011
Game Tree Search Deterministic / stochastic games Perfect / imperfect information games
Finite zero-sum games deterministicchance perfect informationchess, checkers, go, othello backgammon, monopoly, roulette imperfect information battleship, kriegspiel, rock- paper-scissors bridge, poker, scrabble
Game trees
Classical algorithms MiniMax O(w d ) Alpha-Beta O(w d/2 ) √√√ΧΧ√√√ΧΧ max min max
Advanced search techniques Transposition tables Time efficiency / high cost of space PVS Negascout NegaC* SSS* / DUAL* MTD(f)
Uncertain Reasoning O(w d/2 ) More cut-offs <5<5 ?≥5 <5<5 √√ΧΧΧ√Χ√ΧΧ max min max
Game tree statistical evaluation Minimax value Tree count
Game tree analytical evaluation FXFX FXFX FXFX FXFX F min F max Probability density Cumulative distribution
Game tree analytical evaluation FXFX FXFX FXFX FXFX F min F max
Cumulative probability function by level
Probability density function by level
Relative performance (Leaf nodes visited)
Hey! That's My Fish!
Evaluation function Fish Amount (player) – Fish Amount (opponent)
Iterative deepening
Number of positions searched
Relative number of positions searched
Relative time elapsed
Conclusions and Future Work BNS gives a 10 percent performance improvement Transposition tables Different evaluation functions Multi-player game Approximation search
Questions ?