Private Transformer Inference in MLaaS: A Survey

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Yang, Zhou, Xinyu, Wang, Yitong, Qian, Liangxin, Zhao, Jun
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908365362298880
author Li, Yang
Zhou, Xinyu
Wang, Yitong
Qian, Liangxin
Zhao, Jun
author_facet Li, Yang
Zhou, Xinyu
Wang, Yitong
Qian, Liangxin
Zhao, Jun
contents Transformer models have revolutionized AI, powering applications like content generation and sentiment analysis. However, their deployment in Machine Learning as a Service (MLaaS) raises significant privacy concerns, primarily due to the centralized processing of sensitive user data. Private Transformer Inference (PTI) offers a solution by utilizing cryptographic techniques such as secure multi-party computation and homomorphic encryption, enabling inference while preserving both user data and model privacy. This paper reviews recent PTI advancements, highlighting state-of-the-art solutions and challenges. We also introduce a structured taxonomy and evaluation framework for PTI, focusing on balancing resource efficiency with privacy and bridging the gap between high-performance inference and data privacy.
format Preprint
id arxiv_https___arxiv_org_abs_2505_10315
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Private Transformer Inference in MLaaS: A Survey
Li, Yang
Zhou, Xinyu
Wang, Yitong
Qian, Liangxin
Zhao, Jun
Cryptography and Security
Artificial Intelligence
Transformer models have revolutionized AI, powering applications like content generation and sentiment analysis. However, their deployment in Machine Learning as a Service (MLaaS) raises significant privacy concerns, primarily due to the centralized processing of sensitive user data. Private Transformer Inference (PTI) offers a solution by utilizing cryptographic techniques such as secure multi-party computation and homomorphic encryption, enabling inference while preserving both user data and model privacy. This paper reviews recent PTI advancements, highlighting state-of-the-art solutions and challenges. We also introduce a structured taxonomy and evaluation framework for PTI, focusing on balancing resource efficiency with privacy and bridging the gap between high-performance inference and data privacy.
title Private Transformer Inference in MLaaS: A Survey
topic Cryptography and Security
Artificial Intelligence
url https://arxiv.org/abs/2505.10315