Hello everyone, I am a novice. I have known that ensemble and ensemble can establish a connection, but what I want to know is, is there any connection between neurons in an ensemble and neurons? Or to put it another way, how do neurons in an ensemble make connections with neurons?
Hi @chenweiwei and welcome! Yes you can have neuron-to-neuron connections. Could you take a look at the “Direct connections” part of the in-depth guide on Nengo connections? Let us know if you have any follow-up questions.
Thank you very much for your reply. I would like to ask you another question. The 10 neurons in the figure are all Izhikevich models with the same parameters. If I want to make some changes, for example, the parameters of 3 neurons are 0.02, 0.25, -65, 8, the parameters of the other 3 neurons are 0.02, 0.2, -65, 8, and the parameters of the remaining 4 neurons are 0.2, 0.3, -65, 8. So what do I do?
That’s a great question! Usually we think of all of the neurons in a single ensemble as having the same parameters controlling the neuron models (e.g., membrane time-constant, refractory period, the Izhikevich parameters, etc.) but I can see the reason for wanting this. You might want to try copying the Izhikevich class from neurons.py and modifying it to allow numpy arrays for those 4 parameters. I think all of the math should just work out due to the way numpy will broadcast operations elementwise across these arrays, but not sure. If you are newer to Python or not comfortable with trying this yourself, let one of us know.
You might also find it easier to just create 3 different ensembles (where each set of neurons within an ensemble have the same parameters).
You could put those within a Network with input/output nodes that would allow you to connect to and from those neurons as if they were a single object, if that is important (see EnsembleArrays for a similar idea).
2 Likes
Hello @arvoelke ,
Now I am a bit of confusion the connection weight in Nengo. As you said, that means in decoded connection, the neurons in pre-ensemble don’t connect each neuron in post-ensemble, right?
@ntquyen2, to answer your question:
No, they do not. I’ve elaborated more in your other thread, here: LIF neuron activity - #26 by xchoo