DeepSMOTE

This repository provides a unoffical pytorch implementation of "DeepSMOTE: Fusing Deep Learning and SMOTE for Imbalanced Data"

Prerequsits

conda env create -f environment.yml
conda activate DeepSMOTE

demo of results

naive autoencoder

left: origional image right: reconstruction	sythetic images using smote

autoencoder with random permutation

left: origional image right: reconstruction	sythetic images using smote

Run the code

## train model
python main.py --train

## test model (show reconstructions and sythetic images)
python main.py

To Do

dataset

network archicture

hand-craft CNN
UNet

🌟 Please star this repo if you find it helpful!

If you find this git repo useful, please consider citing the associated paper and acknowledge our implementation using the snippet below:

@article{peng2022imbalanced,
  title={Imbalanced Classification in Medical Imaging},
  author={Peng, Le and Travadi, Yash and Zhang, Rui and Cui, Ying and Sun, Ju},
  journal={arXiv preprint arXiv:2210.12234},
  year={2022}
}

PL97/DeepSMOTE