"A stochastic, iterative, evolutionary general purpose search strategy based on the principles of population genetics and natural selection".
The genetic algorithm was proposed by J.W. HOLLAND (1975, 1992) as a way to the simulation of adaptive population systems. He generalized it as "genetic operators" models that can be used for the study of optimization problems and more recently for automata learning.
The genetic algorithm, not being narrowly deterministic, does not lead to just a simple solution, but on the contrary opens the way for a progressive and adaptive search for better solutions in evolving conditions.
HOLLAND distinguishes the following transforming operations: crossing-over, inversion, mutation, selection.
These operations are found in nature. But they are frequent in any system wherein numerous agents act collectively as a population engaged in an adaptive search.
A. AGAPIE writes: "Genetic algorithms (GAS) are robust probabilistic algorithms for optimization, relying strongly on parallel computation. their power comes from multi-point exploiting of the searching space"(2000, p. 35)
→ Neural networks; Parallel distributed processing