Repo for the AML exam material.
Source of the original code https://github.com/eric-moreno/Anomaly-Detection-Autoencoder
The used dataset is synthetic and contains both realistic LIGO noise events and injection events, in which a GW waveform is superimposed on the synthetic instrumental noise. It has been generated with the program available at this repo https://github.com/timothygebhard/ggwd
Link to a minimal (working, used for training on Colab) dataset: https://drive.google.com/file/d/18raqTXI-BLZYGVggGUc6pcQ9NggtEykK/view?usp=sharing
train.ipynb and eval.ipynb contain the code (and a few comments) to train and test the AEs, respectively. (NB: train.ipynb is blank since the training has been performed on Colab. Here the link to one of the used notebook https://colab.research.google.com/drive/11bZVRvpJskp8AUb-ncRuBQTIcfC8GQgY?usp=sharing)
The output directory contains some output examples.