Code used for GECCO 2022 Paper: "Evolving SimGANs to Improve Abnormal Electrocardiogram Classification" (wksp229s2)
$ conda activate simgan-cgp
$ python main.py -p problem_simgan_ecg
$ python main.py -p problem_simgan_transform
UPDATE May 23, 2022
After creating this tag, we realized that there wasn't a convenient way to get the ECG data used in our SimGAN, so we compiled the code we used and made it into a single script.
Here is the python file to download and create the npy files we used:
https://github.com/ezCGP/ezCGP/blob/2022S-stable/data/download_data/simgan_ecg.py
Be sure to read the documentation at the top on additional python packages that need to be installed.
Additionally, change lines 122 + 123 in data/data_tools/simganData.py to
self.real_raw = np.expand_dims(np.load('./data/datasets/ecg/abnormal.npy', allow_pickle=True), axis=1)[:32]
self.simulated_raw = np.expand_dims(np.load('./data/datasets/ecg/sim_dataset.npy', allow_pickle=True), axis=1)[:400]
Or if you are git savvy, just checkout the following commit hash e1dc2e80691082da652625ed63c776c680d98758.