Interesting genetic programming exercise

[bruto]

I don't quite know the relevance of this, but seeing a thread relating to the evolution of vision reminded me of this site I recently ran across.

http://rogeralsing.com/2008/12/07/genetic-programming-evolution-of-mona-lisa/

[CapelDodger]

Wow. Way better than Life :).

[Paul C. Anagnostopoulos]

How is this rule implemented?

"4) If the new painting looks more like the source image than the previous painting did, then overwrite the current DNA with the new DNA."

~~ Paul

[BenBurch]

How does one run it/install it? I tried the binary he provided under XP and it errored-out.

[ThatSoundAgain]

In the discussion at the site, some people are accusing the algorithm of really being a stochastic hill climber. Author rebuts that it's a genetic algorithm with a population of one.

Either way, if this could be sped up a bit, you'd have some neat image compression.

ETA: Pop. size of two, now - parent and child. It's changed since I read it.

[GodMark2]

In the discussion at the site, some people are accusing the algorithm of really being a stochastic hill climber. Author rebuts that it's a genetic algorithm with a population of one.

Either way, if this could be sped up a bit, you'd have some neat image compression.

ETA: Pop. size of two, now - parent and child. It's changed since I read it.

The parent and child never combine their 'DNA', so they never form a 'population'. As such, it's not a genetic algorithm. It simply says "Did that last change make things better or worse?" Simple stochastic hill-climber.

Which doesn't stop it from being pretty cool.