Genetic Programming in Machine Language
Overview
Genetic programming is a technique for evolving computer programs aimed at solving specific problems. At any given time, a population of such programs is maintained, following a dynamics inspired from the evolution of real animals. That is, the better a given program solves a problem, the greater will be his chances of surviving and passing "genetic" information to the next generation. If two programs are good enough to be selected for "mating", two offspring programs are then created, each consisting of instructions taken from the parents.
Specifics of Genetic Programming in Machine Language
The classical approach to genetic programming is to evolve an interpretable representation of a program (such as a tree or a sequence of high-level instruction). In contrast to this, the machine language approach uses machine code as program encoding, which allows for direct, uninterpreted execution, on a target processor.
My contribution
I have created a genetic programming system which works on Intel 32-bit processors, using either integer or floating point operations. When polished, this project will be published under GNU licences. If you just can't wait, send me an email.