(New page: Genetic Algorithm is inspired from the evolutionary process and is on method which can be used when there are sufficient resources for computation. In GA based classification we will init...) |
|||
| Line 11: | Line 11: | ||
The algorithm itsef is motivated by human evolutionary process of survival of the fittest. We can summarize the GA to have the following steps: | The algorithm itsef is motivated by human evolutionary process of survival of the fittest. We can summarize the GA to have the following steps: | ||
| − | + | 1.Choose the initial population | |
| − | + | 2.Assess the fitness of each member | |
| − | + | 3.Select the best members to form the off spring | |
| − | + | 1. Breed new genration through cross over and mutation | |
| − | + | 2. Evaluate fitness of the offspring and replace defecient ones Repeat step 3 until we have the best solution | |
GA is widely used as a optimization technique. Other biological inspired techniques include Ant Colony Optimization and Particle Swarm Optimization. | GA is widely used as a optimization technique. Other biological inspired techniques include Ant Colony Optimization and Particle Swarm Optimization. | ||
Sources: DHS | Sources: DHS | ||
| − | + | Wikipedia | |
Revision as of 23:41, 21 March 2008
Genetic Algorithm is inspired from the evolutionary process and is on method which can be used when there are sufficient resources for computation.
In GA based classification we will initialize several classifiers (called initiapopulation or chromosomes) . Once we have done this we evaluate the fitness of each classifier. We could do this by observing how they classify data whcih belong to some known pre labelled classes. This can be done using a "fitness" function. Once the fitness has been assessed we generate off spring classifiers in a stochastic manner .It may occur that some offspring classifiers have higher fitness than their parent some will have lower. Basically the best ones are retained and randomly the next generation is produced.
Because of the ranking (fitness) each generation is generally better than its previous.
GA may take long time to converge though.
GA can be parallelized easily and if there are facilities available for Parallel Processing it can be a tool to be considered.
The algorithm itsef is motivated by human evolutionary process of survival of the fittest. We can summarize the GA to have the following steps:
1.Choose the initial population
2.Assess the fitness of each member
3.Select the best members to form the off spring
1. Breed new genration through cross over and mutation
2. Evaluate fitness of the offspring and replace defecient ones Repeat step 3 until we have the best solution
GA is widely used as a optimization technique. Other biological inspired techniques include Ant Colony Optimization and Particle Swarm Optimization.
Sources: DHS Wikipedia
