redhat-et / homomorphic-learning Goto Github PK

TenSEAL is a library for doing homomorphic encryption operations on tensors, built on top of Microsoft SEAL. It provides ease of use through a Python API, while preserving efficiency by implementing most of its operations using C++.

As a part of KR2, we would like to explore this tool and create a Helloworld notebook for the same.
Acceptance Criteria -

• A jupyter notebook that uses TenSEAL and performs basic encryption schemes and operations

• Add an OWNERS file
• Add team to the OWNERS file

It might be worth exploring the HE techniques in the order of the history and further understanding why one approach might be better than the others.
We should also add this to the summary document.

Reference

Create a hello world python notebook demonstrating Fully homomorphic encryption.

Assume two stakeholders who wish to keep their pieces of data private, but federate them into a single feature set for training a model, and then deploying it. Synthetic data is acceptable but some kind of actual open data use case is even better.

Perform a literature survey on homomorphic learning (survey papers, for example) to understand the state of the art from an abstract technique perspective.

Acceptance Criteria :

• Find relevant papers, resources, videos, etc related to homomorphic learning
• Select what papers are important for in-depth study and split amongst @suppathak and @aakankshaduggal
• #7

• Add basic timing functionality to TenSEAL notebook for BFV scheme - #22
• Add basic timing functionality for CKKS scheme

Create a README.md file describing the project.

Create a use case demo by taking a simple example dataset and using the encryption technique on it.

As a part of #26 , Try out TenSEAL for training and evaluating a logistic regression model on encrypted data using homomorphic encryption.

• #28
• #29
• #30

Top 3 tools selected so far -

• PyFHEL-(by @suppathak) https://github.com/redhat-et/homomorphic-learning/blob/main/notebooks/pyfhel_hello_world.ipynb
• SEAL-python- [PR review pending] - #17
• TenSEAL - [PR review pending - https://github.com//pull/22, https://github.com//pull/23

Since redhat-et org doesn't have any cookie-cutter project template, we feel we should import the template just for this Data Science Project, just to get a headstart and a better organized repository.

cc - @suppathak @MichaelClifford @erikerlandson