The advancement of large language models (LLMs) has significantly enhanced the ability to
effectively tackle various downstream NLP tasks and unify these tasks into generative …

Abstract Secure 2-party computation (2PC) enables secure inference that offers protection
for both proprietary machine learning (ML) models and sensitive inputs to them. However …

ChatGPT is recognized as a significant revolution in the field of artificial intelligence, but it
raises serious concerns regarding user privacy, as the data submitted by users may contain …

Large transformer-based models have realized state-of-the-art performance on lots of real-
world tasks such as natural language processing and computer vision. However, with the …

Secure transformer inference has emerged as a prominent research topic following the
proliferation of ChatGPT. Existing solutions are typically interactive, involving substantial …

With the growing use of large language models hosted on cloud platforms to offer inference
services, privacy concerns are escalating, especially concerning sensitive data like …

Secure two-party neural network (2P-NN) inference allows the server with a neural network
model and the client with inputs to perform neural network inference without revealing their …

We present Rhombus, a new secure matrix-vector multiplication (MVM) protocol in the semi-
honest two-party setting, which is able to be seamlessly integrated into existing privacy …

Many inference services based on large language models (LLMs) pose a privacy concern,
either revealing user prompts to the service or the proprietary weights to the user. Secure …

Homomorphic encryption (HE)-based deep neural network (DNN) inference protects data
and model privacy but suffers from significant computation overhead. We observe …