|
|
Computers > Artificial Intelligence > Genetic Programming > Applications > Genetic Programming and the Game of Go
Genetic Programming and the Game of Go
Submitted 2005-10-07 08:34:26 by psiolent
Rating: 0 (0 votes) | 1578 Views
Section: Applications
I've come across a fairly interesting application of genetic programming. The author has developed a fairly sophisticated and robust framework for genetic programming and has used it to evolve programs that play the ancient game of Go.
The author summarizes his project, named "Lithos", as follows:
Lithos is a stack based evolutionary computation system. Unlike most EC systems, its representation language is computationally complete, while also being faster and more compact than the S-expressions used in genetic programming. The version presented here applies the system to the game of Go, but can be changed to other problems by simply plugging in a different evaluation function.
The page gives a very clear yet detailed explanation of how the framework works as well as how he applied it to Go. Also, he provides source code and binaries so you can run his program yourself.
He describes the application to Go as follows:
The current version of Lithos evolves programs to play the game of Go. It should be emphasized that the objective of this is not to produce something that can defeat a skilled human player - evolved programs under current hardware play very poorly even compared to hand-written programs. (Unsurprisingly, considering that they consist of a few hundred to a few thousand bytes of code, versus tens to hundreds of kilobytes for hand-written programs.) The objective is to experiment with artificial evolution on a problem that - unlike most of those studied, but like biological evolution - is effectively open-ended.
A very interesting read and a cool program to run on your own computer. Check it out. |
Discuss this submission in the forum
|
|
|