search

neurodata / ProgLearn

NeuroData's package for exploring and using progressive learning algorithms
https://proglearn.neurodata.io
Other
33 stars 42 forks source link

Transform data with any valid projection matrix #418

Open eigenvivek opened 3 years ago

eigenvivek commented 3 years ago

Is your feature request related to a problem? Please describe. Supersedes #417

Describe the solution you'd like We're planning a bunch of new MORF experiments with different types of patches (convolutional filters, gabor filters, etc.)

Allowing the user to define and pass their own projection matrix sampling functions would make these experiments a lot easier, rather than having a new splitter class for every patch sampling method we want to try

I'm proposing modifying ObliqueSplitter's sample_proj_matrix to allow for user-defined matrix sampling methods.

Alternatively, i could make a new splitter class with this functionality

@levinwil @jdey4 @PSSF23 what do y'all think?

PSSF23 commented 3 years ago

I personally think that modifying the existing function makes more sense, so it could save us an additional splitter class?