Off Topic > Off Topic
Making a new evolution simulator.
Meldaril:
Are you using a steady-state genetic algorithm approach? Some other heuristic approach, a different EA?
What sort of crossover (edge recombination) and are you using mutation as well?
How are you performing fitness evaluation on a given set of candidate solutions?
DrenDran:
--- Quote from: Meldaril on February 05, 2017, 10:35:51 AM ---How are you performing fitness evaluation on a given set of candidate solutions?
--- End quote ---
No fitness function. Those that breed the most will spread their genes the most.
--- Quote from: Meldaril on February 05, 2017, 10:35:51 AM ---What sort of crossover (edge recombination) and are you using mutation as well?
--- End quote ---
There will be both loveual and aloveual reproduction.
For the aloveual reproduction there's a few major different types of mutation:
[o]Deletion of segments
[o]Duplication of segments
[o]Insertion of random data
[o]Changing of individual values
--- Quote from: Meldaril on February 05, 2017, 10:35:51 AM ---Are you using a steady-state genetic algorithm approach? Some other heuristic approach, a different EA?
--- End quote ---
There's no generations per-se since as above the creatures simply reproduce when their code tells them to.
Juncoph:
--- Quote from: Ipquarx on February 04, 2017, 11:43:30 PM ---This is actually a really roundabout way to bugtest your physics engine. Whatever creature you make will evolve to exploit any flaw in the system for its advantage.
--- End quote ---
simulated evolution is also potentially how self-improving ai could become dangerous
DrenDran:
New video's up.
Evolution can actually happen now, I'll post an update with any interesting evolution tomorrow.
Nonnel:
this is dope. looks fun to play around with.