GEMS InfoServer

The GEMS InfoServer contained herein exists as a forum for the promulgation of background and technical information on the author's investigations and results in the field of Genetic Programming (GP). Specifically, this InfoServer reports on research and results involving the genetic evolution of machine-language software.

ABSTRACT

GP, of which GEMS is one instantiation, is an approach to answering a key artificial intelligence question in computer science, to wit, "How can computers learn to program themselves to solve problems, especially problems whose solutions vary stochastically over time."

GEMS is a process that automatically produces machine-language software with a desired functional capability.

GEMS uses the techniques of genetic algorithms (GA) and genetic programming (GP) to search the solution space of machine language programs to identify one or more programs that satisfy the problem goal.

A technical paper entitled "Genetic Evolution of Machine Language Software" in Microsoft word format can be downloaded here.

The remainder of this InfoServer is broken into topical sections as follows.

GENETIC ALGORITHMS -- GP & GEMS use Genetic Algorithms (GAs)

GENETIC PROGRAMMING BACKGROUND -- Prior work in the field of GP

WHY MACHINE LANGUAGE (ML) -- Contrasting ML vs prior GP implementations

A GEMS ARCHITECTURE -- A specific architecture for GEMS research

THE `HELLO WORLD' PROBLEM -- Initial parameterization of the GEMS process

GEMS TEST RESULTS IN STRING GENERATION -- Parametric Data

Go to Arktist Home Page

Last Updated: August 28, 2008