Efficient Homomorphic Integer Computer from CKKS

Описание: Speaker: Jaehyung Kim, Stanford University
Host: Patrick Longa, Microsoft Research Redmond

Fully homomorphic encryption (FHE) has evolved from Gentry’s original blueprint into a diverse family of practical schemes, including BGV/BFV for exact arithmetic, DM/CGGI-style schemes for fast binary computation, and CKKS for high-throughput approximate arithmetic. I will begin with a brief overview of this evolution and the main ideas that shaped modern FHE. The focus of the talk is my recent work on extending CKKS beyond approximate numerical computation to support reliable and efficient discrete and integer arithmetic. I will introduce the discrete CKKS framework, which reformulates CKKS so that encrypted data can behave like exact integers or bits, while preserving the efficiency and structure that make CKKS attractive in practice. Building on this, I will show how CKKS can be used as a practical engine for general-purpose integer computation. In simple terms, this makes it possible to run programs that manipulate large integers directly on encrypted data, even though CKKS was originally designed for approximate arithmetic. Such programs include cryptographic, database, and systems-level computations. These constructions demonstrate that CKKS can serve as a unified platform for both numerical and discrete computation, significantly broadening its scope for privacy-preserving systems.

See more at https://www.microsoft.com/en-us/resea...