"Specify initial weights to EvolutionaryWeighting"
Is there any way to specify an initial set of weights to EvolutionaryWeighting? Given how long the algorithm takes to run, I'd like to be able to take an initial guess or rough approximation from another approach that can be found more quickly, and give it to RM to get it nearer to the optimal solution to help get it started.
Another example of how this could be useful would be to run, say, 25 generations now, to get an approximate answer that is workable when I need an initial unrefined answer, but later (not necessarily part of the same RM session), restart the EvolutionaryWeighting where it left off for another 25 generations to improve the fit, get more refined weight estimates, and continue iteratively towards the ideal solution.
As documented, EvolutionaryWeighting doesn't seem to allow you to specify an initial guess to work from. I'm wondering if the RM gurus have any tricks up their sleeves to get something like this to work.
Thanks, as always,
Keith
Another example of how this could be useful would be to run, say, 25 generations now, to get an approximate answer that is workable when I need an initial unrefined answer, but later (not necessarily part of the same RM session), restart the EvolutionaryWeighting where it left off for another 25 generations to improve the fit, get more refined weight estimates, and continue iteratively towards the ideal solution.
As documented, EvolutionaryWeighting doesn't seem to allow you to specify an initial guess to work from. I'm wondering if the RM gurus have any tricks up their sleeves to get something like this to work.
Thanks, as always,
Keith
Find more posts tagged with
Sort by:
1 - 1 of
11
hey, interesting idea. Never thought about this. This is currently not possible but I have written it down to our todo list.
Cheers,
Ingo