C++ for Z

Overview

helayers-lab-s390x is based on HELayers which is written in C++ and includes a C++ API that enables data scientists and application developers to easily use the power of FHE by supporting a wide array of analytics such as Linear/Logistic Regression and Neural Networks. HELayers has been designed with a layered set of capabilities and coupled with appropriate APIs so users can fully utilize the services provided by the SDK. HELayers is delivered as an open platform, capable of using the latest FHE implementations for a given use case. It is enabled with patented optimization and performance boosting innovation for computation, AI innovation and use case requirements that facilitate the practical use of a wide variety of AI workloads over FHE data.

System Requirements

HElayers is a Linux based Docker image. The only requirement is a system that is capable of running a container, such as Docker, with the necessary user privileges to run docker commands. It is also assumed you have a working internet connection.

This image is built to run on all zSystems architectures including Linux on IBM Z (z16 and LinuxOne III), and zCX Container Extensions

License

This image is provided under a community edition license for non-commercial use; see license. For commercial deployments and access to the source code, please contact chamliam@ie.ibm.com for the premium edition license.

Call to Action

To help us create the best possible offering tailored to your needs, please provide feedback by taking the HElayers Experience Survey
For more information on FHE, please visit our Content Solutions Page
For more information on IBM HElayers, please visit our IBM Developer Blog

Installation

Note: Do not expose the HElayers containers to unauthorized access. Anyone with access to the container web interface will be able to execute code inside it. Only start the HElayers containers on local personal computers or on servers with controlled access.

Step 1: Pull the Docker image

In your terminal pull this image from IBM Container Registry to your local repository.

docker pull icr.io/helayers/helayers-lab-s390x:latest

If you specify :latest, it will always pull the most recent version. However, if you are looking for a specific version you can replace :latest with the version number, such as :1.0.

If the fetch is successful, it will download the image and you will now have an image in your local repository named icr.io/helayers/helayers-lab:latest. Check that this image exists with

docker images

Step 2: Running the image

Once the image download is complete, run the image as a container in Docker...

docker run -p 8443:8443 -d --rm --name helayers-lab-s390x icr.io/helayers/helayers-lab-s390x:latest

This command runs the image and sets the container to be accessed via port 8443. It also gives the container a name called helayers-lab-s390x, which we can use to find it in the list of other running containers.

Step 3: Access HElayers

Open a web browser on your host machine (not the docker container instance), and browse to http://127.0.0.1:8443/ . This will connect you to the welcome page of the VS Code.

Note: if you run the web browser in different machine than the docker container, then you need to change the IP address of the URL.

Step 4: Getting Started with the tutorials

Click on the application menu (three bars icon, top left), and Choose Terminal > New Terminal

In the terminal change directory to examples/

Follow the directions in the readme file of that demo.

Available demos:

Hebase tutorial

Basic layer 1 (hebase - abstract layer) tutorial. It demonstrates helayer's low level API for manipulating ciphertexts directly.

Neural network tutorials

Step by step tutorials how to use the C++ API for NN inference

Linear regression

Compute linear regression using encrypted model and data.

Logistic Regression Inference on a Credit Card Fraud Detection Dataset

Build a logistic regression model encrypted under HE, and run inference of encrypted samples from a dataset of credit card transactions.

Nearest neighbor

Encrypt a set of centroids and find nearest neighbor under HE. Given an encrypted samples, we compute the distance between each sample and each centroid under encryption. On the client side, the results are decrypted and automatically post-processed to obtain the nearest neighbor.

Bitwise tutorial

Tutorial explaining the bitwise API (implemented with the BGV scheme)

Decicion tree inference

Decision tree inference for credit card fraud detection

Neural network demos

Neural network inference for MNIST, medical, and credit-card fraud datasets. Note that the neural network tutorials above are more recommended for beginners.

BGV world country db lookup

Encrypted query over an encrypted database. Uses the BGV scheme, and Fermat's little theorem to compute equality over the modular arithmetic supplied by the scheme.