[comp.ai.neural-nets] Speeding up genetic algorithms

kingsley@hpwrce.HP.COM (Kingsley Morse) (01/05/91)

reiser@pmafire.inel.gov (Steve Reiser) writes in sci.med:
 
>In tests reported in a 1989 issue of "Noetic Science Review", it was
>found that a bacteria exposed to a virus which is fatal to the bacteria 
>had a 70% survival rate due to a genetic double mutation which required
>mutation "A" to occur before mutation "B" at specific locations on the
>DNA.  The mutations protected the bacteria from the virus.
>
>The article suggested that much of evolution may have been non-random
>and selective mutations occur allowing species to advance at an infinitely
>faster rate than would be probable by random genetic mutation.
>
>This finding suggests that evolution is harder to explain away than it
>had been when statistical probabilities of evolution creating man was
>used to suggest that it was necessary for a god to intervene.

This seems to imply that *genetic algorithms* might be speed up if mutations
aren't random. Has anyone heard of "selective mutations" and how they work?