Further to my discussion on CUDA in my previous post, it seems that some work has been done before in using CUDA for hand evaluation.
More information can be found here:
http://www.ii.metu.edu.tr/coursewebsites/quda/cuda_zone/report.htm
In this report, CUDA was used to compare all possible hand combinations providing performance improvements. However, in my work I am using a large random sample of hands to provide a reasonably accurate, but not exact, result. If anything, using a sample should enable a computer poker player to play more like a human since humans often don't know the exact strength of their hand, using a upper and lower bound instead.
If anything, using a sample cash games should enable a computer poker player to play more like a human since humans often don't know the exact strength.
ReplyDeleteThat's true, using example human behaviour should enable a computer poker player to play more like a human, but in my project I wanted to avoid introducing examples of human intelligence to see if it is possible for intelligence to emerge by itself.
Delete