Set of experiments reported in the paper "Optimality Conditions and a Trust Region Algorithm for Bilevel Parameter Learning in Total Variation Image Denoising".
This module depends on Tuomo's Valkonen AlgTools and ImageTools packages as well as VariationalImaging and TestDatasets developed by David Villacís.
$ hg clone https://tuomov.iki.fi/repos/AlgTools/
$ hg clone https://tuomov.iki.fi/repos/ImageTools/
Once cloned the repositories, we need to upload those to the julia package manager
pkg> develop AlgTools
pkg> develop ImageTools
pkg> add https://github.com/dvillacis/VariationalImaging.git
pkg> add TestDatasets
To reproduce the experiments it is necessary to clone the code repository and initialize the julia modules
$ git clone https://github.com/dvillacis/BPLDenoising.git
$ julia --project=BPLDenoising
Once in the julia REPL just import the module and the experiment functions
julia> using BPLDenoising
julia> scalar_bilevel_tv_learn("dataset_name")
For the "dataset_name" variable you can choose one from the TestDatasets package.