Concrete-ML is a Privacy-Preserving Machine Learning (PPML) open-source set of tools built on top of The Concrete Framework by Zama. It aims to simplify the use of fully homomorphic encryption (FHE) for data scientists to help them automatically turn machine learning models into their homomorphic equivalent. Concrete-ML was designed with ease-of-use in mind, so that data scientists can use it without knowledge of cryptography. Notably, the Concrete-ML model classes are similar to those in scikit-learn and it is also possible to convert PyTorch models to FHE.
Data scientists can use models with APIs which are close to the frameworks they use, with additional options to run inferences in FHE.
Concrete-ML features:
- built-in models, which are ready-to-use FHE-friendly models with a user interface that is equivalent to their the scikit-learn and XGBoost counterparts
- support for customs models that can use quantization aware training. These are developed by the user using pytorch or keras/tensorflow and are imported into Concrete-ML through ONNX
Depending on your OS, Concrete-ML may be installed with Docker or with pip:
| OS / HW | Available on Docker | Available on pip |
|---|---|---|
| Linux | Yes | Yes |
| Windows | Yes | Coming soon |
| Windows Subsystem for Linux | Yes | Yes |
| macOS (Intel) | Yes | Yes |
| macOS (Apple Silicon, ie M1, M2 etc) | Yes | Coming soon |
Note: Concrete-ML only supports Python 3.7 (linux only), 3.8 and 3.9.
Concrete-ML can be installed on Kaggle (see question on community for more details), but not on Google Colab (see question on community for more details).
To install with Docker, pull the concrete-ml image as follows:
docker pull zamafhe/concrete-ml:latest
To install Concrete-ML from PyPi, run the following:
pip install -U pip wheel setuptools
pip install concrete-ml
You can find more detailed installation instructions in this part of the documentation
A simple example which is very close to scikit-learn is as follows, for a logistic regression :
from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split
from concrete.ml.sklearn import LogisticRegression
# Lets create a synthetic data-set
x, y = make_classification(n_samples=100, class_sep=2, n_features=30, random_state=42)
# Split the data-set into a train and test set
X_train, X_test, y_train, y_test = train_test_split(
x, y, test_size=0.2, random_state=42
)
# Now we train in the clear and quantize the weights
model = LogisticRegression(n_bits=8)
model.fit(X_train, y_train)
# We can simulate the predictions in the clear
y_pred_clear = model.predict(X_test)
# We then compile on a representative set
model.compile(X_train)
# Finally we run the inference on encrypted inputs !
y_pred_fhe = model.predict(X_test, execute_in_fhe=True)
print("In clear :", y_pred_clear)
print("In FHE :", y_pred_fhe)
print(f"Similarity: {int((y_pred_fhe == y_pred_clear).mean()*100)}%")
# Output:
# In clear : [0 0 0 0 1 0 1 0 1 1 0 0 1 0 0 1 1 1 0 0]
# In FHE : [0 0 0 0 1 0 1 0 1 1 0 0 1 0 0 1 1 1 0 0]
# Similarity: 100%This example is explained in more detail in the linear model documentation. Concrete-ML built-in models have APIs that are almost identical to their scikit-learn counterparts. It is also possible to convert PyTorch networks to FHE with the Concrete-ML conversion APIs. Please refer to the linear models, tree-based models and neural networks documentation for more examples, showing the scikit-learn-like API of the built-in models.
Full, comprehensive documentation is available here: https://docs.zama.ai/concrete-ml.
Various tutorials are proposed for the built-in models and for deep learning. In addition, several complete use-cases are explored:
-
MNIST: a python script and notebook showing quantization-aware training following FHE constraints. A fully-connected neural network is implemented with Brevitas and is converted to FHE with Concrete-ML.
-
Titanic: a notebook, which gives a solution to the Kaggle Titanic competition. Implemented with XGBoost from Concrete-ML, this example comes as a companion of the Kaggle notebook, and was the subject of a blogpost in KDnuggets.
-
Sentiment analysis with transformers: a gradio demo which predicts if a tweet / short message is positive, negative or neutral, with FHE of course! The live interactive demo is available on Hugging Face. This blog post explains how this demo works!
-
CIFAR10 FHE-friendly model with Brevitas: code for training from scratch a VGG-like FHE-compatible neural network using Brevitas, and a script to run the neural network in FHE. FHE simulation shows an accuracy of 88.7%, but running inference with FHE is still a work-in-progress.
-
CIFAR10 / CIFAR100 FHE-friendly models with Transfer Learning approach: series of three notebooks, that show how to convert a pre-trained FP32 VGG11 neural network into a quantized model using Brevitas. The model is fine-tuned on the CIFAR data-sets, converted for FHE execution with Concrete-ML and evaluated using FHE simulation. For CIFAR10 and CIFAR100, respectively, our simulations show an accuracy of 90.2% and 68.2%. True FHE inference is a work-in-progress.
-
FHE neural network splitting for client/server deployment: we explain how to split a computationally-intensive neural network model in two parts. First, we execute the first part on the client side in the clear, and the output of this step is encrypted. Next, to complete the computation, the second part of the model is evaluated with FHE. This tutorial also shows the impact of FHE speed/accuracy tradeoff on CIFAR10, limiting PBS to 8-bit, and thus achieving 62% accuracy.
-
Encrypted image filtering: finally, the live demo for our 6-min is available, in the form of a gradio application. We take encrypted images, and apply some filters (for example black-and-white, ridge detection, or your own filter).
More generally, if you have built awesome projects using Concrete-ML, feel free to let us know and we'll link to it!
To cite Concrete-ML, notably in academic papers, please use the following entry, which list authors by order of first commit:
@Misc{ConcreteML,
title={Concrete-{ML}: a Privacy-Preserving Machine Learning Library using Fully Homomorphic Encryption for Data Scientists},
author={Arthur Meyre and Benoit Chevallier-Mames and Jordan Frery and Andrei Stoian and Roman Bredehoft and Luis Montero and Celia Kherfallah},
year={2022-*},
note={\url{https://github.com/zama-ai/concrete-ml}},
}
This software is distributed under the BSD-3-Clause-Clear license. If you have any questions, please contact us at hello@zama.ai.