Ai in an Artificial life simulation

[subcosin]

im planning on making an alife simulation with artificial lifes(with artificial brains) ,an artificial world ,
weather , different plants/herbs that contain poison/cures ,etc...

i would like this simulation to consist of the following:

* lifes that learn to find food
* lifes that learn to find a mate for reproduction
* lifes that can contract viruses and also cure them
... and more

I think i have most of it figured out exept for the AI part .I have a
little experience in AI ,my main questions are:

* What type of neural network would be used for these lifes?
* What would be fed to the inputs of these neural networks?
* What would i do with the output of these neural networks

i assume it should be unsupervised?
i looked at `reinforcement learning` but dont know if this is the
right answer,There seems to be that many different types of neural
networks and training sets that it has all got me a bit confused

Its seems that all the tutorials ive found on neural networks are all
based around charcter recognition and stock market prediction,etc...

Thanks for any replies

[zameer]

Well,

You need to know about some stuffs related to Agent Based Systems and Multi Agent Technology, which are some kind of Agent societies. The concept is some kind of fully automation.

In technical terms the entire system is based on a Blackboard style architecture. (Remind Publisher-Subscriber design pattern). In this case in a Multi agent system each Agent can update it's state to other agents via a common knowledge (black board) base called common ontology. And other Agents can update it self based on the status of that Agent.

It is like 2 way of Publisher-Subscriber design pattern. (i.e Publisher becomes a subscriber and vice versa)

Then the entire system system will evolve untill they reach a final solution.

There are many Multi Agent Simulation kits available in the Internet. You can try them. But better to read some stuffs before you begin any work related to MAS.

Hope AForge would introduce such a framework in future.

Thanks.
zameer

[techy12]

That's a great idea but I think that would require lots of studies on important matters that would be needed in the process.
Human minds are really great when it comes to innovations and technology. Another interesting about technology is about the human and computer hybrids that have always been a theoretical possibility, but not an actual likelihood. Now, two new developments bring biological computing closer than ever. The two brand new improvements in creating human-computer hybrids are getting this closer. Biological transistors and skin-based touchpads have now been developed and even tested. These developments and others have been making the possibility of biologically compatible computers more likely. Medical science is especially interested in the possibility, as programmable, biologically neutral computers would be able to improve diagnostics. Medical science is not the only use for these developments, but it is sure to be one of the first uses. Truly usable, commercial biological computers are likely to be years, if not decades away. These developments, however, are two major steps toward the reality.