fp_datasets
Description
All experiments described in "Closed-Form Feedback-Free Learning with Forward Projection" are reproduced by running the python script provided in "forward_projection/setup/full_script_06_01_25". PTBXL data is provided in: Wagner, P. et al. PTB-XL, a large publicly available electrocardiography dataset. Scientific Data 7 (2020). URL http://dx.doi.org/10.1038/s41597-020-0495-6. Human NonTATA promoters data is provided in: Greˇsov´a, K., Martinek, V., ˇCech´ak, D., ˇSimeˇcek, P. & Alexiou, P. Genomic benchmarks: a collection of datasets for genomic sequence classification. BMC Genomic Data 24 (2023). URL http://dx.doi.org/10.1186/s12863-023-01123-8. OCT and CXR datasets are available in: Kermany, D. S. et al. Identifying medical diagnoses and treatable diseases by image-based deep learning. Cell 172, 1122–1131.e9 (2018). URL http://dx.doi.org/10.1016/j.cell.2018.02.010. NPZ files are loaded in the form (X_trainval, Y_trainval, X_test, Y_test, folds), where the folds are a 5-fold cross-validation split of the rows of X_trainval and Y_trainval. ECG annotations are provided for the first 15 MI-positive ECGs in the test partition of the PTBXL-MI data. Version 2 uploaded due to incomplete uploads of CIFAR10 and CXR datasets in Version 1.
Files
Steps to reproduce
Preprocessing code provided at github.com/robertoshea/forward_projection