The design of competitive algorithms via genetic algorithms

Genetic algorithms have recently become a popular artificial technique for solving complex optimization problems and a sophisticated tool for machine learning. In this paper, we demonstrate the capacity of genetic algorithms to deal with a challenging task, namely the design of competitive algorithms (or heuristics) that manipulate and control dynamic systems for online processing. Dynamic systems that must be controlled by an online heuristic ore very common in computer science applications and the difficulty to find heuristics for these types of systems as illustrated by the enormous efforts in the literature to design good heuristics. We initiate the use of genetic algorithms to aid the design of heuristics to control online dynamic systems by obtaining heuristics for self-organizing lists.<>