Kaleidoscope: Learnable Structured Linear Maps 🤖

A spotlight video describing the paper "Kaleidoscope: An Efficient, Learnable Representation For All Structured Linear Maps", appearing at the International Conference on Learning Representations (ICLR) 2020.

We propose a differentiable family of "kaleidoscope matrices," prove that all structured matrices can be represented in this form, and use them to replace hand-crafted linear maps in deep learning models.

Paper: https://openreview.net/forum?id=BkgrBgSYDS
Blogpost: https://dawn.cs.stanford.edu/2019/06/13/butterfly