by using multiobjective genetic algorithm we can obtain more than one solution but i read some paper that have serveral solution but using only Simple genetic algorithm is it possible and how, when I change Crossover parameter and mutation parameter I get more than one solutions is there another way?
Hi,
This is why in many cases, it is possible to obtain better results by simple GA, However, it depends on defining the fitness function well (not operators). It is suggested that more than 2 objectives are converted to a single objective by using appropriate weighting.
Multi-objective GA is very powerful tool, but it should be used in cases that a compromising between different objectives is required (if there are some independent variables)
With single-objective optimization, you get a (single best) solution that optimizes your single objective.
With multiobjective optimization, you can get a set of solutions that are equally fit in the multi-objective criterion, but have different fitnesses for the individual objectives, which defines the Pareto frontier, see http://en.wikipedia.org/wiki/Pareto_efficiency#Pareto_frontier Solutions on this frontier can subsequently be used for compromising between different objectives, as Ali suggests.
Hi
When you combine several objective functions in a single one you have to use predetermined weight parameters. If you change these weights you can get multiple solutions.
You can refer to the following paper
https://www.researchgate.net/publication/3450508_Multiobjective_Evolution_of_Neural_Controllers_and_Task_Complexity
Article Multiobjective Evolution of Neural Controllers and Task Complexity
- Badges
- Science topic
- Similar topics
- Engineering
- Petroleum Engineering
- Oil