The history of genetic algorithms

In 1859 Charles Darwin published the revolutionary book, The Origin of Species, which governed the way in which species evolved to form better members by the process of natural selection. It was based upon work in the Galapagos Islands on the prominent features of birds and other creatures that lived there and had not always lived there. Genetic research as a scientific domain increased with technology and genetic algorithms(GA) research originated from the work of John Holland and his colleagues at the University of Michigan in the 1970s. Indeed the book published by Holland in 1975 was considered to be the start of GA research. It is based on population statistics of Fisher in the 1950's. However, much of this research was theoretical and it was not until the 1980s that any real applications could be produced from the research. Since De Jong and Hollstein's work in the early 1980s there has been many applications in GA's across many disciplines, including optimisation and route tracking.

The algorithm

genetic algorithms(GA) are a type of stochastic search method derived from an analogy to genes and their study. Like simulated annealing (SA), neural networks(NN) and ant colony optimisation(ACO), GA's are another type of artificially intelligent optimisation which derive from natural biological systems. The metaphor comes from the basis for the algorithm. It is essentially the generation of an artificial population of chromosomes related to our own DNA strings. This population is mapped to solutions to the problem at hand and the population naturally selects the fittest members of its population for mutation or crossover to produce child populations. This is achieved by applying selection pressure to the population much akin to the Darwinian theory of natural selection. These child populations gradually become more and more 'fit' and so converge upon an optimal solution to the problem at hand. The difference between GA's and other optimisation methods are that:
  • GA's search solution populations not a single point
  • Ga's use an objective coding of the populations parameters not the parameters themselves
  • GA's are probabilistic



Bookmark and Share

Submit a comment about this page












I am available for freelance web design work in the Cambridge, UK area, outside normal working hours. If you are interested, please