Re:Genetic Algorithm 2D Car Thingy
by sevensor on Friday November 20, 2015 @12:37PM
(#10602491)
Measuring evolutionary algorithm performance is a complicated question. It depends on what you want out of your algorithm. This one does a magnificent job at what it's for, which is illustrating what artificial evolution looks like.
For an engineering optimization problem, I always look for sources of conflict, so there would be multiple objectives, we'd have to choose a metric or three, do a bunch of different random seed trials, and so forth.
For an engineering optimization problem, I always look for sources of conflict, so there would be multiple objectives, we'd have to choose a metric or three, do a bunch of different random seed trials, and so forth.