Re:HTML5 Genetic Cars
by grinich on Tuesday June 25, 2013 @06:55PM
(#5942992)
Actually, the thing genetic algorithms try to do is break out of this local maximum, mutating past a trait correlated with poor performance in hopes of finding a more fit solution.
[ Parent
| on HN ]
In my hypothetcial algorithm, of course mutation would still take place.
But my current instace of the web page is at a point where it's wasting a lot of time on unfit 'mutts' and it seems like I could get to a "best rhinoboat" and "best assdragger" solution faster if it would stop wasting time exploring the mediocre blend.
Yeah, most effective genetic algorithms take a hybrid approach where they only do more dramatic mutations for some of the population, and then optimize the rest with a regular fitness function and small variations. The goal with these systems is to have several different but effective breeds emerge.
I didn't dig into this implementation, but it seems like it doesn't carry many (more than one?) individuals between cycles.