Generative SDRs

Hello everyone,

I recently made a small post on the generative properties of SDRs when combined with certain training algorithms. It’s based on K-sparse autoencoders. A blog post is available here:

It’s not directly HTM related, but it does involve SDRs (binary sparse codes). Perhaps it is useful for HTM in some way as well.


GSDR has already been ported to Python and available via pip install gsdr! Code and several example python notebooks here.

1 Like