Fighting games and Go

Adaptive AI for Fighting Games

Traditionally, AI research for games has focused on developing static strategies—fixed maps from the game state to a set of actions—which maximize the probability of victory. This works well for discovering facts about the game itself, and can be very successfully applied to combinatorial games like chess and checkers, where the personality and play style of the opponent takes a backseat to the...

Fighting m-criminals with Bayesian Stackelberg games

Bayesian Stackelberg games have been used to compute optimal patrol strategies for the security of the Los Angeles International Airport (LAX) by modelling the security situation as a 1-leader, 1-follower, Bayesian game where there is one leader type and an arbitrary number of follower types. A previous approach, named DOBSS, computes the Stackelberg solution for this game by solving a Mixed In...

A Googolplex of Go Games

Reinforcement Learning of Intelligent Characters in Fighting Action Games

Abstract. In this paper, we investigate reinforcement learning (RL) of intelligent characters, based on neural network technology, for fighting action games. RL can be either on-policy or off-policy. We apply both schemes to tabula rasa learning and adaptation. The experimental results show that (1) in tabula rasa leaning, off-policy RL outperforms on-policy RL, but (2) in adaptation, on-policy...

Fighting poverty, fighting AIDS.

J effrey sachs is an economist, not an evangelist. But give the man 10 minutes behind a microphone, and a scholarly symposium starts to feel like a revival meeting. “Ladies and gentlemen,” he tells a hushed hall after describing how AIDS sufferers die in Malawi for want of $1-a-day drugs, “this plague is exploding. Its consequences will make the world quake. Rich countries could stop the devast...