Cryptographic inference for large language model via secret sharing

Inference services based on large language models may lead to the leakage of user input hints to the server or proprietary model weights to the user. Cryptographic techniques such as secure multi-party computation and homomorphic encryption provide feasible solutions to the above problems, but they...

Cijeli opis

Bibliografski detalji

Glavni autori:	CHENG Ke, XIA Yuheng, DAI Chuanyun, FU Jiaxuan, ZHU Xinghui, SHEN Yulong
Format:	Artigo
Jezik:	Chinês
Izdano:	Editorial Department of Journal on Communications 2025-06-01
Serija:	Tongxin xuebao
Teme:	privacy-preserving large language model secret sharing secure multi-party computation cryptographic inference
Online pristup:	http://www.joconline.com.cn/zh/article/doi/10.11959/j.issn.1000-436x.2025115/
Oznake:	Dodaj oznaku Bez oznaka, Budi prvi tko označuje ovaj zapis!