ScatNet (v0.2)

ScatNet (Scattering Network) is a MATLAB software which includes scattering transforms of one-dimensional signals and images, with applications to classifications of audio signals, multifractals and images. It includes demos which reproduce the figures of our publications. ScatNet is a pointer to tutorials to teach how to perform the scattering transform.

Mailing list

You can fill in the following form to get the latest updates.

Softwares released by the team

Name Latest version Type Date of release Primary authors Reference papers Notes
ScatNet 0.2 Audio, Image November 2013 L. Sifre, J. Anden [Andén and Mallat, 2011b] , [Sifre and Mallat, 2013]
ScatNetLight 0.1 Image May 2015 E. Oyallon [Oyallon and Mallat, 2015] Requires to be an advanced user of ScatNet
NanoScat 1 Image October 2015 C.E. Cella - Minimalist version of the scattering transform (e.g. for pedagogical purposes)
HaarScat 1 Graph December 2014 X. Cheng, X. Chen (Mia) [Chen et al., 2014]
Phase recovery 1 Reconstruction June 2012 I. Waldspurger [d'Aspremont et al., 2015]
Compact LIBSVM 1 Classifier June 2013 J. Anden - 32 bits precision, triangular kernels, multicore and in-place. Speed up calculations and reduce memory usage.


Note that these packages are provided for legacy purposes. However, links and articles can be missing. To compute scattering transforms, please use the ScatNet toolbox above.