[Abstract]

Owing to its capabilities for fixed-point arithmetic and SIMD-like vectorization, RNS-CKKS stands out among fully homomorphic encryption (FHE) schemes for enabling computations on encrypted data, making it a popular choice for privacy-preserving machine learning services. While previous efforts have made strides toward automating the challenging task of scale management essential for RNS-CKKS’s fixed-point arithmetic, they have overlooked the significance of performance, output errors, and compile time. This oversight restricts the ability of users to investigate and optimize the trade-off between error margins and latency. I will present a comprehensive approach to optimizing FHE applications through advanced compiler frameworks, scale management schemes, and performance analysis techniques, namely Hecate, ELASM, and Reserve Analysis. They not only demonstrate the feasibility of efficient, privacy-preserving applications but also open new avenues for further research in optimizing encrypted computation. Additionally, I will introduce an automatic bootstrap placement compiler called DaCapo, which allows to scale privacy-preserving applications.

 

 

[Biography]

Yongwoo Lee received the B.S. degree in Computer Science and Engineering from POSTECH in 2019 and the Ph.D. degree in Electrical and Electronic Engineering from Yonsei University in 2024. He subsequently worked as a postdoctoral researcher at Yonsei University. He is currently an Assistant Professor with the Department of Electrical Engineering and Computer Science at DGIST. His research interests include compiler and runtime systems, with a particular focus on homomorphic encryption.