TechPrufer
Get started
Google
Google logoGoogle · Feature trail

Fully Homomorphic Encryption

1 post · Jun 2021 – Aug 2023

What this trail covers

Google continues to advance Fully Homomorphic Encryption (FHE) by releasing new tools and expanding its toolkit. This includes a TensorFlow-to-FHE compiler for private machine learning on encrypted data, the Jaxite software library for running FHE on hardware accelerators like GPUs and TPUs, and HEIR (Homomorphic Encryption Intermediate Representation), an open-source compiler toolchain for FHE interoperability. These advancements aim to lower costs, expand use cases to video processing and text

All Google trails
  1. Expanding our Fully Homomorphic Encryption offering- Google Developers BlogThis post introduces a TensorFlow-to-FHE compiler that allows trained TensorFlow ML models to be compiled into FHE versions for private inference on encrypted data. It also announces Jaxite, a software library for running FHE programs on hardware accelerators (GPUs, TPUs) by building on JAX. Finally, it introduces HEIR, an open-source compiler toolchain for FHE designed for interoperability across schemes, compilers, and hardware, built on MLIR.
  2. Our latest updates on Fully Homomorphic Encryption- Google Developers BlogThis post announces the release of a TensorFlow-to-FHE compiler, the Jaxite software library for running FHE on hardware accelerators (GPUs, TPUs), and HEIR (Homomorphic Encryption Intermediate Representation), an open-source compiler toolchain for FHE interoperability. These advancements aim to lower costs and expand FHE use cases to video processing and text.