Pyfhel: PYthon for Homomorphic Encryption Libraries

Abstract

Fully Homomorphic Encryption (FHE) allows private computation over encrypted data, disclosing neither the inputs, intermediate values nor results. Thanks to recent advances, FHE has become feasible for a wide range of applications, resulting in an explosion of interest in the topic and ground-breaking real-world deployments. Given the increasing presence of FHE beyond the core academic community, there is increasing demand for easier access to FHE for wider audiences. Efficient implementations of FHE schemes are mostly written in high-performance languages like C++, posing a high entry barrier to novice users. We need to bring FHE to the (higher-level) languages and ecosystems non-experts are already familiar with, such as Python, the de-facto standard language of data science and machine learning. We achieve this through wrapping existing FHE implementations in Python, providing one-click installation and convenience in addition to a significantly higher-level API. In contrast to other similar works, Pyfhel goes beyond merely exposing the underlying API, adding a carefully designed abstraction layer that feels at home in Python. In this paper, we present Pyfhel, introduce its design and usage, and highlight how its unique support for accessing low-level features through a high-level API makes it an ideal teaching tool for lectures on FHE.