A library for Partially Homomorphic Encryption in Python
Clone or download
wilko77 Merge pull request #69 from n1analytics/learning_rate_fix_for_federat…
…ed_learning

make learning rate independent of dataset size
Latest commit fe24881 Sep 11, 2018

README.rst

python-paillier Latest released version on PyPi

CI status of Master
Documentation Status
Requirements Status of master

A Python 3 library for Partially Homomorphic Encryption.

The homomorphic properties of the paillier crypto system are:

  • Encrypted numbers can be multiplied by a non encrypted scalar.
  • Encrypted numbers can be added together.
  • Encrypted numbers can be added to non encrypted scalars.

Running unit tests

python setup.py test

Or use nose:

nosetests

Note related to gmpy2

gmpy2 is not required to use the library, but is preferred. A pure Python implementation is available but gmpy2 drastically improves performances. As indication on a laptop not dedicated to benchmarking, running the example examples/federated_learning_with_encryption.py provided in the library took: - 4.5s with gmpy2 installed - 35.7s without gmpy2 installed

However, gmpy2 is a requirement to run the tests.

Code History

Developed at Data61 | CSIRO.

Parts derived from the Apache licensed Google project: https://code.google.com/p/encrypted-bigquery-client/

No audit

This code has neither been written nor vetted by any sort of crypto expert. The crypto parts are mercifully short, however.