Is there a canonical model or heuristic for the relative importance of mutation versus recombination contingent upon X where X is some small set of intuitive factors? I have a fuzzy argument forming, and maybe that would help me think about it.
-
-
As a cstheorist, I'm contractually obliged to be a downer. I think that the consensus among cstheorists (not GA folks) is that, in general, recombination tends to help only on easy problems (or problems specifically designed to be easy for recombination & hard for local search)
-
On most practical hard problems, either there is a smart combinatorial optimization algorithm that exploits some hidden orded or almost nothing works (and in that case simple local searches tend to outperform recombination like systems).
- 4 more replies
New conversation -
-
-
I mean, you kinda got to it in the previous tweet, it seems like you want some evolution models? like population genetic models or something?
-
Yea. Problem is the search space isn't subject to experimentation or even much measurement, I don't think. So the argument I want to make is much more qualitative and guiding more than anything else.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.