VESTA: A Secure and Efficient FHE-based Three-Party Vectorized Evaluation System for Tree Aggregation Models

  • Haosong Zhao
  • , Junhao Huang
  • , Zihang Chen
  • , Kunxiong Zhu
  • , Donglong Chen
  • , Zhuoran Ji
  • , Hongyuan Liu

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

Machine Learning as a Service (MLaaS) platforms facilitate collaborative machine learning, but trust issues necessitate privacy-preserving methods. As for tree ensembles, prior Fully Homomorphic Encryption (FHE)-based approaches suffer from slow inference speed and high memory usage. This work proposes the VESTA system that leverages compile-time precomputation and parallelizable model partitioning to enhance performance. VESTA achieves a 2.1x speedup and reduces memory consumption by 59.4% compared to state-of-the-art solutions.

Original languageEnglish
Title of host publicationSIGMETRICS Abstracts 2025 - Abstracts of the 2025 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems
Pages31-33
Number of pages3
ISBN (Electronic)9798400715938
DOIs
StatePublished - 9 Jun 2025
Event51st ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems, SIGMETRICS Abstracts 2025 - Stony Brook, United States
Duration: 9 Jun 202513 Jun 2025

Publication series

NameSIGMETRICS Abstracts 2025 - Abstracts of the 2025 ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems

Conference

Conference51st ACM SIGMETRICS International Conference on Measurement and Modeling of Computer Systems, SIGMETRICS Abstracts 2025
Country/TerritoryUnited States
CityStony Brook
Period9/06/2513/06/25

Keywords

  • domain-specific compiler
  • homomorphic encryption
  • parallelism
  • tree ensembles
  • vectorization

Fingerprint

Dive into the research topics of 'VESTA: A Secure and Efficient FHE-based Three-Party Vectorized Evaluation System for Tree Aggregation Models'. Together they form a unique fingerprint.

Cite this