1
$\begingroup$

I've read some about genetic algorithms and the general approach, but I haven't found anything about using it when the length of the solution is unknown. How would the generation of the initial population look like? Would mutation then include adding/removing alleless? Are there any general methods for this?

$\endgroup$
  • $\begingroup$ In general, the answers are all yes. But you haven't provided much information to work with here. You want definitely to able to "hit" all possible genes, so sometimes adding must be necessary. Often you can choose to encode the genes as fix length, e.g. in travelling salesman or subset problems. $\endgroup$ – Pål GD Feb 20 '18 at 18:54
  • $\begingroup$ @PålGD So is there any general approach or should I just think logically? In my case there is a game where you have to find a solution in a finite number of moves, but it could be 1 move or much more. $\endgroup$ – Ferus Feb 20 '18 at 19:03
  • $\begingroup$ Does every gene have a well-defined fitness? You have to balance between generating any possible gene, or only feasible genes. $\endgroup$ – Pål GD Feb 20 '18 at 19:14
  • 1
    $\begingroup$ I don't think there are any general approaches, you will have to try different strategies to see what works. You can relax the problem/game by removing rules and see which mutations/crossovers gives you best fitness. $\endgroup$ – Pål GD Feb 20 '18 at 19:15
  • $\begingroup$ Depending also on your score landscape, there are frameworks that take into account covariance and expected improvements, but I'm not sure that would be possible in your situation. $\endgroup$ – Pål GD Feb 20 '18 at 19:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.