https://github.com/diffeqml/deep-signal-processing

Accompanying notebook guides for the deep signal processing notes [TBA].
https://github.com/diffeqml/deep-signal-processing

convolution deep-learning fourier-transform frequency-domain long-sequence signal-processing state-space-models transfer-functions

Last synced: 7 days ago
JSON representation

Accompanying notebook guides for the deep signal processing notes [TBA].

• Host: GitHub
• URL: https://github.com/diffeqml/deep-signal-processing
• Owner: DiffEqML
• License: apache-2.0
• Created: 2022-11-24T06:20:21.000Z (about 2 years ago)
• Default Branch: main
• Last Pushed: 2022-11-24T07:20:03.000Z (about 2 years ago)
• Last Synced: 2024-11-07T06:32:20.717Z (about 2 months ago)
• Topics: convolution, deep-learning, fourier-transform, frequency-domain, long-sequence, signal-processing, state-space-models, transfer-functions
• Language: Jupyter Notebook
• Homepage:
• Size: 1.51 MB
• Stars: 4
• Watchers: 2
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
Deep Signal Processing

 Accompanying notebooks for [TBA] 


## The Atlas of Convolutions 

### Part 1: Memory, Causality and Parameter Scaling

* `basics`: introduces the basic idea of a linear convolution and the different ways of implementing it

* `fft_conv`: discusses diagonalization of circulant matrices, motivating an efficient method to convolve signals

* `causality`: investigates how to enforce causality of a convolution 

* `ssm_kernel`: provides a showcase of a simple diagonal state space and the resulting kernel

* `transfer_function`: primer on transfer functions, properties and how to parametrize a convolution as a ratio of polynomials over the complex numbers

* `analysis`: explains how to leverage amplitude and phase of a frequency response to inspect input-output pairs for pure sinusoidal signals

* `parametrizations`: provides a set of minimal `nn.Module` classes implementing the different convolution parametrizations introduced in the notes.

## Other excellent resources

State spaces: 

* [state-spaces (repo)](https://github.com/HazyResearch/state-spaces).

* [Simplifying S4 (blog)](https://hazyresearch.stanford.edu/blog/2022-06-11-simplifying-s4).