N-dimensional data

[ziofil]

Hello, I'd like to try your code on a dataset that has about 400 examples and 10 features per example.

I seem to have understood that your code assumes data with 2 features? Or can it work on data with N features?

[djstrouse]

Hey, are the features continuous or discrete? Since this is an implementation of the exact information bottleneck, and not say a variational version, the dataset needs to be pretty small, since you need to enumerate the joint distribution of inputs and outputs P(x,y). For continuous features, this requires binning. Also, what dimensionality is your output? Or are you clustering?

…

On Fri, Jun 28, 2019 at 10:55 AM ziofil ***@***.***> wrote: Hello, I'd like to try your code on a dataset that has about 400 examples and 10 features per example. I seem to have understood that your code assumes data with 2 features? Or can it work on data with N features? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2?email_source=notifications&email_token=AAB3U2T64R6DUPMDHEJQBWDP4XN23A5CNFSM4H4D5PIKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4G4I4HEA>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAB3U2R5MXPOE55SNK5SWP3P4XN23ANCNFSM4H4D5PIA> .

[ziofil]

4 discrete and 6 continuous, but I can bin them. The output is binary, it’s a binary classification in the end.
Now I understand the meaning of pxy 😆