Event based version of rate based homeostasis

[schmitts]

Hi!

Cf. https://github.com/brian-team/brian2/issues/1241; does anybody know about comparisons of event based versions of rate based homeostasis?

Thanks,

Sebastian

[kernfel]

I have nothing to add myself, I just want to avoid your question falling to the wayside due to the broken link. Thus, for completeness:
Fixed link to the issue

@schmitts

Hi!

I’m looking for an example for homeostatic plasticity to stabilize firing rates. Is there a notion for firing rates available that can be used to formulate a weight update rule?

Thanks,

Sebastian

@mstimberg

In spiking networks, there is no direct firing rate variable that is available. What you’d typically do is to use a variable that increases with each spike (via the reset statement) and decays over time. Another option is to low-pass filter the membrane potential and to use this as a proxy for the firing rate (see e.g. the Clopath et al. (2010) example).

[schmitts]

Thanks for the heads up!

[mstimberg]

Hi again. Not sure this is exactly what you are looking for, but Zenke et al. (2013) implemented a homoeostatic mechanism with spikes that calculates the firing rate as a low-pass filtered version of the spike rate, like I mentioned in the github issue.

[schmitts]

Hi!

I’ve put together an example for spike (event) based homeostasis: Add example for spike based homeostasis by schmitts · Pull Request #1331 · brian-team/brian2 · GitHub

Cheers,

Sebastian

[schmitts]

It got merged, thanks! Example: spike_based_homeostasis — Brian 2 2.4.2.post0.dev296 documentation