Noise robustness project #58
thesamovar
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I have an undergraduate student Ziduan Li who will be working on this project over the summer, looking at the effect of adding noise. I'm away in August so if anyone is able to provide guidance for her if she has questions, that would be great. Brief outline of project below.
The idea is to add noise to the model and look at the effect on performance. Here's how I was thinking of doing it. At the moment, we generate spikes like this:
This gives us a time varying firing rate FR(t) which we then convert into a time varying Poisson spike train like this:
This gives us spike trains like this:
Now we will add a noise term like so:
Giving code like this:
And spike trains like this:
I think the signal to noise ratio would be this (but check):
And we might expect to see a performance curve that looks something like this:
I've highlighted here that this curve may depend on the other parameters, for example a higher value of the envelope power might lead to better performance (or it might not, not sure).
Ziduan will work on generating plots like this, and then investigating what causes performance to go up or down and which parameters make it more or less robust to noise.
Beta Was this translation helpful? Give feedback.
All reactions