Dopamine: Differentially Private Secure Federated Learning on Medical Data

• A submission to ITU AI/ML in 5G Challenge for ITU-ML-5G-PS-022

• For following future work on this project please see the main repo in the following link: https://github.com/ipc-lab/private-ml-for-health

While rich medical datasets are hosted in hospitals distributed across countries, concerns on patients' privacy is a barrier against utilizing such data to train deep neural networks (DNNs) for medical diagnostics. We propose Dopamine, a system to train DNNs on distributed medical data, which employs federated learning (FL) with differentially-private stochastic gradient descent (DPSGD), and, in combination with secure multi-party aggregation, can establish a better privacy-utility trade-off than the existing approaches. Results on a diabetic retinopathy (DR) task show that Dopamine provides a privacy guarantee close to the centralized training counterpart, while achieving a better classification accuracy than FL with parallel differential privacy where DPSGD is applied without coordination.

Folders (Submission to ITU challenge):

1. report: includes the final report. For 1.Design document showing the reasons for the choice of privacy-preserving technique and the network architectural components.
2. private_training: includes the source code and a JupyterNotebook tutorial for training the privacy-preserving model explained in the report. For 2.Source code for the implementation of the privacy-preserving design across various architectural components.
3. private_inference: includes the source code and demo for running inference on the privately trained model. For 3.Tested code and Test Report for all implementations- Implementations of Privacy-Preserving AI Technique, Trained Data Model, UI on smartphone.
4. video_demo: include some video demos showing how to run training and inference. For 4. A Video of the demonstration of Proof-of-Concept.

Tutorial

We provided a Jupyter Notebook for training on Google Colab. Please see the file JNotebook_running_FSCDP_on_Colab.ipynb in the private_training folder.

Live Demo:

Please use this link to get an inference on a Diabetic Retinopathy medical image:

https://imperial-diagnostics.herokuapp.com/

(Note: implementing the pure private inference is still in progress...)

Preprint

Please find the most recent preprint of this project in the following link:

https://github.com/ipc-lab/private-ml-for-health/blob/main/private_training/Dopamine.pdf

Citation

If you find the provided code or the proposed algorithms useful, please cite this work as:

@article{dopamine2020,
  title={Dopamine: Differentially Private Secure Federated Learning on Medical Data},
  author={Mohammad Malekzadeh, Burak Hasircioglu, Nitish Mital, Kunal Katarya, Mehmet Emre Ozfatura, Deniz Gündüz},  
  url = {https://github.com/ipc-lab/private-ml-for-health}
  year={2020}
}