课程: AI Algorithms for Game Design with Python

免费学习该课程！

今天就开通帐号，24,700 门业界名师课程任您挑！

Monte Carlo evaluation functions

Monte Carlo evaluation functions - Python教程

课程: AI Algorithms for Game Design with Python

开始一个月试用 购买团队方案

Monte Carlo evaluation functions

“

- [Instructor] Let's go back to the limited depth problem. By now, you should know that evaluation functions are tricky to get right. That's because we don't really have a clue of the actual quality of our functions. These functions are usually based on our own speculation. So here's a clever idea, which is sometimes called a Monte Carlo evaluation function. Suppose we are at the maximum limited depth and we need an evaluation value. Instead of calling an arbitrary function, we could keep on playing the game by making a random move, then another and another, another, and so on until we reach a terminal node. In this example, it's a winning scenario of value plus 100, but we don't have to stop there. We could play another sequence of random moves for both players until we get a terminal node again. This example shows a tie, that's a value of zero. And so on, we could do this several times. This time, we got a losing outcome of minus 100. In this other sequence, we got minus 100 again…

内容