Kymatio: Scattering Transforms in Python

In plain words: Kymatio is a Python package that computes the wavelet scattering transform, turning 1D, 2D, or 3D signals into features that stay steady when the signal shifts or warps. On graphics cards it runs much faster than CPU versions and uses little memory.

Abstract

The wavelet scattering transform is an invariant signal representation suitable for many signal processing and machine learning applications. We present the Kymatio software package, an easy-to-use, high-performance Python implementation of the scattering transform in 1D, 2D, and 3D that is compatible with modern deep learning frameworks. All transforms may be executed on a GPU (in addition to CPU), offering a considerable speed up over CPU implementations. The package also has a small memory footprint, resulting inefficient memory usage. The source code, documentation, and examples are available undera BSD license at https://www.kymat.io/

Mathieu Andreux, Tomás Angles, Georgios Exarchakis, Roberto Leonarduzzi, Gaspar Rochette, Louis Thiry, John Zarka, Stéphane Mallat, Joakim andén, Eugene Belilovsky, Joan Bruna, Vincent Lostanlen, et al.
arXiv:1812.11214 · cs.LG, cs.CV, cs.SD, eess.AS, stat.ML · submitted Dec 28, 2018 · updated May 31, 2022
abstract · pdf · html