Genetic Algorithm - Structure

<< previous

answermath HOME

answermath en Español

next >>

Artificial Intelligence Tutorials: Data Mining ▫ Neural Networks ▫ Fuzzy Logic ▫ Genetic Algorithms

Tutoriales de Inteligencia artificial: Redes Neuronales ▫ Lógica Fuzzy ▫ Minería Datos ▫ Algoritmos Genéticos

Mental math Tips: Addition ▫ Subtraction ▫ Multiplication ▫ Division ▫ Sine ▫ Cosine ▫ Tangent

Tips cálculo mental: Suma ▫ Resta ▫ Multiplicación ▫ División ▫ Seno ▫ Coseno ▫ Tangente

Structure

Aside from some important parameters, the central idea of the algorithm is this:

1. At any given time we will have a set (a Generation) of possible and different solutions of the problem (individuals, or phenotypes) with their respective abstract representation (chromosomes, or genotype).

2. An evaluation of each solution based on how good it is, creates some kind of ranking. This “Fitness” quantification of the solutions is one of the most important parts of the algorithm.

3. We may decide to keep as members for the next generation some of the best individuals in the present one (Elitism).

4. A new generation of solutions (Offspring) is created with individuals that are obtained by the combination (Crossover) of two selected solutions of the previous generation (Parents).

5. Little changes to few randomly selected solutions provide alternative mechanisms to obtain new individuals (Mutation).

6. Once the new generation of solutions is available we need to evaluate them as in point # 2 in order to begin another cycle. The process will terminate as in many recursive routines: because time is out, no evolution is observed, or an acceptable solution is found.