Evolutionary Algorithms

by machines’ help → the artificial neural networks simulate the brain (Flying vehicles, DNA computers, membrane-based computers)

by algorithms’ help → Evolutionary algorithms simulate the evolution of nature, Particle Swarm Optimisation simulates the collective and social behaviour, Ant Colony Optimisation

Characteristics

• Iterative and parallel processes
• Based on random search
• Bio-inspired – involve mechanisms as: Natural selection, Reproduction, Recombination, Mutation

Jean Baptise de Lamark

• Needs of an organism determine the evolving characteristics
• Useful characteristics could be transferred to offspring

Charles Darwin

• Variation
  • Overproduction of offspring
• Natural selection
  • Competition (generation of constant size)
  • Fitness survival
  • Reproduction
  • Occurrence of new species

Design

Chromosome representation

• External level → phenotype
  • Individual – original object in the context of the problem
  • The possible solutions are evaluated here
  • ant, knapsack, elephant, towns...
• Internal level → genotype
  • Chromosome – code associated to an object
    • Composed by genes, located in loci (fix positions) and having some values (alleles)
  • The possible solutions are searched here
  • One-dimensional vector (with numbers, bits, characters), matrix, …

must be representative for:

• Problem
• Fitness function
• Genetic operators