I was refered to this group by my peers over in comp.robotics.misc. I have read some of the FAQ and I have to admit that some of the math involved is over my head. Basically, I'm working on a machine intelligence experiment for robotic applications. Does anyone have any information on a computer model for a neuron? I am somewhat new to the subject but I have put in alot of thought about how to make this work. It goes like this:
You have a neuron datatype. The datatype has values for input, input weights, output destinations, output weights, and memory storage. The datatype itself would be encapsulated into an object type with methods. Since each instance of an object is a pointer variable, I thought about having an array of objects and processing the data in the array sequentially, from position 0 through position x. Any changes to neuron data or firing states would be stored in a update buffer until the end of the array sequence was encountered. Then the update buffer would be read and all the neuron data would be updated at that point. Additionally, any flags that got set to create/destroy neurons would also be honored at this time. Then the process repeats.
Does this sound practical to anyone? Comments, hints, suggestions?