FBResNet

Forward Backward ResNet.

This work is a continuation of the work done by Cecile Della Valle in https://github.com/ceciledellavalle/FBResNet.

Installation

1. Install requirements:

   $ pip install -r requirements.txt

2. Install the FBRN model from the FBRN directory:

   $ pip install .

3. The model can be tested by using the example notebook in the notebooks folder or by submitting the jobs bash files:

   $ bash ./jobs/test.sh

Files organization

• Datasets
  • Images: two sets of images from which are extracted 1D signal datasets
  • Signals: dataset constructed from images
• FBRN: contains FBRestNet files
  • info.py: model information. Allows pip installing the FBRN model.
  • main.py: contains FBResNet class, which is the main class including train and test functions
  • model.py: includes the definition of the layers in FBRestNet
  • myfunc.py: useful functions used such as convolution and Abel operators
  • proxop: contains he cardan class used to compute the proximity operator of the logarithmic barrier
    • hyperslab.py: proximity operator for hyperslab constraint
    • hyperscube.py: proximity operator for cubic constraint
• scripts
  • FBRN_click.py: interface model parameters and options through a click comand line interface.
  • fouriermethod.py: 1D reconstruction with Fourier filtering method.
  • lowessmethod.py: 1D reconstruction with Lowess filtering method.
  • optuna_FBRN.py: explore and find the best hyperparameters for the FBRN model.
  • train_utils.py: train, test and plot functions.
  • training_time_opt.py: study FBRN trainig times.
• notebooks: collection of jupyter notebooks for different analysis and studies.
• jobs: bahs files for submitting model training and evaluation with different parameters.
• outputs: contains the predictions and the FBRN weights for different model parameters (Not in Github). You can find the output folder as well as the datasets here!
• Figures