This repository contains the code used to train and evaluate the signal vs background discrimination model for the SABRE experiment.
Currently the model is based on a 1 dimensional convolutional neural network (CNN), but many other approaches can be included easily just by adding a new files in the folder models/.
The data format is a folder containing individual txt files with the numbers representing the waveforms.
Two folders are needed: one for the signal and one for the background.
The data is then divided into training, validation and test sets for significant training and evaluation.
To change the specific setting for the run, multiple settings files in json format can be created in the folder settings/.
Right now, the settings include only the path for the data and the output folder for the results, but more can be added easily.
To run the code, just execute the script with:
source run.sh
where the settings file is hardcoded in the script, or with:
source run.sh -j <settings_file>
or
source run.sh --json_file <settings_file>
where the <settings_file> is the json file with the settings.