genetic algorithm

Terms from Artificial Intelligence: humans at the heart of algorithms

The glossary is being gradually proof checked, but currently has many typos and misspellings.

A genetic algorithm takes inspiration from the proceses of natural selection for biological species. Individuals are defined using a set of parameters, effectively artificial genes – in the biological analogy, the genotype. These are used to assess their fitness based on whatever criteria are considered optimal and any constraints – in the biological analogy, the phenotype. The least fit indivdiuals 'die' (removed from the population) and the fittest selected for 'breeding'; typically this process is stochastic, with fitness being used to weigh the proabability of a random decision as to which die and breed. Those chosen to breed may be subject to random mutation and pairs of individuals brought together and their artificial genes randomly mixed, in a process akin to mating. In some versions only mutation is used so that the procsses is more like the growth of bacteria or asexual reproduction. In some variants the new interbred individuals replace their parents' generation entirely, in others they are added to it (replacing those that died).

Used in Chap. 4: pages 52, 55; Chap. 5: page 64; Chap. 8: pages 104, 110; Chap. 9: pages 118, 122, 123, 124, 125, 126; Chap. 10: page 135; Chap. 11: page 159; Chap. 14: page 212; Chap. 18: page 288; Chap. 21: pages 332, 336; Chap. 24: page 376