ICADCML 2021 Paper A Novel Approach to Encrypt Data using Deep Neural Networks

In this paper we propose a novel approach to Encrpyt data using Deep Neural Networks. We propose an Autoencoder techinque which can sucessfully encrypt and decrypt data. We secure this method using keys by ensembling autoencoders.

Project layout

This is how the project is structured. We also provide Colab Notebooks that can be used to reproduce our results.

├── docs                            # Documentation files built using mkdocs
├── models                          # Models that we trained
├── neural_encryption_networks      # This folder contains all code.
│   ├── __init__.py
│   ├── notebooks                   # Reproducible notebooks.
│   └── src                         # Python scripts.
└── requirements.txt                # To install stuff.

Documentation

For detailed documentation visit here

Colab Notebooks

We provide Colab notebooks to directly play with. The are available in neural_encryption_networks/notebooks folder too.

• Encrypting with Deep Neural Networks

• Ensemble Network for Keys

Runing locally

Install the requirments by running

pip install -r requirements.txt

The code uses Tensorflow 2.4. And is Tested on Python 3.6+

We recommend using virtual environements using Conda or similar to avoid conflicts.

cd neural_encryption_networks/src

This folder contains all the code you need !

Citation

We will provide a Bibtex entry soon. For now people can cite this GitHub repository.