Uncertainty quantification for spectral unmixing in gamma-ray spectrometry

This is the repo for uncertainty quantification methods for gamma-ray spectrometry.

The codes are organized as follows:

• The Code folder contains the source code for the IAE and the uncertainty calculation by Fisher and MCMC:
  • LA when X is known
  • LA when X is unknown
  • MCMC when X is known
  • MCMC when X is unknown
• The Data folder contains the dataset of 96 spectral signatures of 12 radionuclides as a function of steel thickness.
• The Notebooks folder contains a jupyter notebook file for quantify the uncertainty
  • The Models folder contains the pre-trained IAE model.

Package requirements

The codes are implemented using Pytorch and Pyro. To do it, you will need the packages listed in environment.yml. To create and activate a conda environment with all the imports needed, do (with CPU):

• conda env create -f environment.yml
• conda activate pytorch
• pip install pyro-ppl==1.8.6

If there is a problem with the installation of Pytorch or Pyro, please follow this link to install it correctly: Pytorch (Cuda or CPU), Pyro

Test the code

• Run Uncertainty_gamma_spectrometry.ipynb if you want to see how to use the codes above to quantify the uncertainty.