An implementation of the generalized scattering transform in Julia. Documentation in progress.
Basic installation: at the moment, this is not an official package. Because of this, to install it, you must first clone it, then make it accessible to your project.
This uses a modified version of Wavelets.jl that can be found here, and added via Pkg.develop("https://github.com/dsweber2/Wavelets.jl.git").
In 0.7 onwards, use either Pkg.add("https://github.com/dsweber2/ScatteringTransform.jl.git"), Pkg.develop("https://github.com/dsweber2/ScatteringTransform.jl.git"), or ] add https://github.com/dsweber2/ScatteringTransform.jl.git in a REPL. Then from a Repl
(v0.7) pkg> activate .
(ScatteringTransform) pkg> instantiate
either way, make sure you don't have more than a single thread when building by
setting JULIA_NUM_THREADS = 1 before calling Pkg.build.
This implementation works strictly on 1 dimensional data and uses Morlet wavelets. If you give it data that has more dimensions, it will transform along the last dimension. There are 2 steps to applying a scattering transform. The first is constructing the transform, done with stParallel(m,example). Then you need to actually transform the data; if you are investigating a single example, use st(f, layers) to get a Scattered, a type containing both the intermediate results and the output of each layer. On the other hand, if you're transforming a large set, and only want the outputs, possibly highly subsampled, use thinSt.
For more detailed description, see the file basicUsage.jl