What type of study is this?

This is a Literature Review study.

October 16, 2025Open Access

Private Transformer Inference in MLaaS: A Survey

YLYang LiNanjing University of Chinese Medicine XZXinyu ZhouNorth Sichuan Medical University YWYitong WangSichuan University

Key Points

Private transformer inference enhances user data privacy during machine learning tasks, addressing concerns with centralized processing.
Utilizing cryptographic methods like homomorphic encryption, private transformer inference allows safe data handling without compromising model functionality.
A structured taxonomy for private transformer inference was developed, providing a framework to evaluate privacy solutions against efficiency requirements.
Balancing high-performance inference with robust data privacy remains a significant challenge in current machine learning applications.

Abstract

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.

Perguntar à IA

Bookmark

View Full Paper