Can NEURON model a "memory" network?

Moderator: wwlytton

Post Reply
Kane
Posts: 7
Joined: Mon Oct 25, 2010 1:41 pm

Can NEURON model a "memory" network?

Post by Kane » Tue Dec 21, 2010 8:13 am

I've built a small network, trying to train it. Is that possible, I don't whether NEURON is a dynamic or not?
Can Neuron support synaptic plasticity or changing the weight adaptively?

ted
Site Admin
Posts: 5566
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Can NEURON model a "memory" network?

Post by ted » Wed Dec 22, 2010 1:16 pm

Kane wrote:I've built a small network, trying to train it. Is that possible
Yes. What learning rule do you have in mind?
I don't whether NEURON is a dynamic or not?
What do you mean by "dynamic"?
Can Neuron support synaptic plasticity or changing the weight adaptively?
If you can describe the rule for plasticity by a formula or algorithm, it can be implemented.

Kane
Posts: 7
Joined: Mon Oct 25, 2010 1:41 pm

Re: Can NEURON model a "memory" network?

Post by Kane » Tue Jan 04, 2011 2:41 pm

ted wrote:
Kane wrote:I've built a small network, trying to train it. Is that possible
Yes. What learning rule do you have in mind?
I don't whether NEURON is a dynamic or not?
What do you mean by "dynamic"?
Can Neuron support synaptic plasticity or changing the weight adaptively?
If you can describe the rule for plasticity by a formula or algorithm, it can be implemented.
dynamic means it's not a time-invariant system.
I'm a neurologist, finding it's too hard to program in NEURON. If NEURON is going to be widely used, it should be much easier to use.
Anyway, thanks for your help, and I will keep on it.

ted
Site Admin
Posts: 5566
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Can NEURON model a "memory" network?

Post by ted » Wed Jan 05, 2011 5:53 pm

Kane wrote:dynamic means it's not a time-invariant system.
If by this you mean "parameters can change in the course of a simulation" the answer is yes, NEURON is "dynamic."
I'm a neurologist, finding it's too hard to program in NEURON.
NEURON does lack built-in learning rules and GUI tools for specifying them. I wonder if NEST or a connectionist simulator would be more to your liking.

Post Reply