Closed JohnnyQAQ closed 3 years ago
Hi @JohnnyQAQ, I am almost sure that they are using the pre-processed data set from Kaggle: https://www.kaggle.com/shayanfazeli/heartbeat. The data set is divided into mitbih_train.csv and mitbih_test.csv. I'm using "almost", because when using the model to generate the ECG signals in the same conditions on Colab, I'm running out of memory.
I would be glad if someone could clarify the out of memory issue.
Here is a link to a shared Google Drive folder with an example of the data we used for this other paper.
Please Note: the data is formatted as multichannel ECG with a shape of [109246,187,2]. If you are having an OOM issue just use a sub set of this data. Also for the code in this repo just select one of the available leads.
Say you want 50K samples then your array shape should be [50000, 187, 1]
Cheers
Hi,I am a college student from China. I want to reproduce it,But I lack the experimental data(mitbih_train.csv and mitbih_test.csv),I hope you can share them,thank you so much!