Python-FHEz

Python-FHEz is a privacy-preserving Fully Homomorphic Encryption (FHE) and deep learning library.

Interactive graph example of an FHE compatible neural network:

0%

Fully Homomorphic Encryption:

Deep Learning:

This library is capable of both fully homomorphically encrypting data and processing encrypted cyphertexts without the private keys, thus completely privately.

This library also supports:

• advanced serialization of cyphertext objects, for transmission over networks using marshmallow

• Kernel-Masqueradeing

• Hadmard Product cyphertext branching

Cite

Either:

@online{fhez,
  author = {George Onoufriou},
  title = {Python-FHEz Source Repository},
  year = {2021},
  url = {https://gitlab.com/DeepCypher/Python-FHEz},
}

Or if you do not have @online support:

@misc{fhez,
  author = {George Onoufriou},
  title = {Python-FHEz Source Repository},
  howpublished = {Github, GitLab},
  year = {2021},
  note = {\url{https://gitlab.com/DeepCypher/Python-FHEz}},
}

Table of Contents

• 1. Documentation Variations
• 2. Source Code Variations
• 3. The Open Software License 3.0 (OSL-3.0)
• 4. Fully Homomorphic Encryption
• 5. Installation
• 6. Examples
• 7. Neural Network
• 8. Traversal
• 9. FHE Parameterisation
• 10. Layers
• 11. Activations
• 12. Optimisers
• 13. Loss Functions
• 14. Operations

Also See

This is part of a larger body of related together work. We are building similar tools for go (DarkLantern). We are also building fully open-source infastructure to run FHE using the server-client model, as well as offering this as a service to others (DeepCypher GitLab, DeepCypher.me)