Getting Started with FHEON

OpenFHE

FHEON is built on OpenFHE. To set up FHEON, first install OpenFHE following the instructions on the OpenFHE website.

FHEON has been tested on OpenFHE v1.2.3, v1.2.4, and v1.3.1.
We recommend Linux or macOS (Ubuntu 22.04 recommended).
FHEON is also compatible with HE Accelerators and has been tested on HEXL Accelerator.

Prerequisites

Before installing FHEON, ensure your system has the following:

FHEON Installation

Clone the FHEON repository:

git clone https://github.com/stamcenter/fheon.git
cd FHEON
mkdir HEkeys
mkdir images **Put all the datasets in this folder**

Create a build directory:
```
mkdir build && cd build
```

Configure and build:

cmake -DCMAKE_BUILD_TYPE=Release ..
make -j$(nproc)

Run examples inference built e.g:
```
./lenet5
```

Basic Neural Network Examples

LeNet-5 FHE Implementation

This example demonstrates the LeNet-5 inference workflow using FHEON. It shows how to configure the FHE context, encode MNIST image data, and run convolution, average pooling, ReLU, and fully connected layers in the encrypted domain.

Full source: LeNet5.cpp

ResNet-20 FHE Implementation

This example demonstrates ResNet-20 inference using FHEON. It shows how to set up FHE context, model weights, residual blocks, and classification layer.

Full source: ResNet20Basic.cpp

For more details and examples, visit the dedicated Examples section in this documentation, where we show different models. We also provide a step by step guide for their full implementation details.

FHEON on GitHub

View FHEON GitHub