{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,3]],"date-time":"2026-07-03T06:17:25Z","timestamp":1783059445407,"version":"3.54.6"},"reference-count":5,"publisher":"IEEE","license":[{"start":{"date-parts":[[2026,3,16]],"date-time":"2026-03-16T00:00:00Z","timestamp":1773619200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,3,16]],"date-time":"2026-03-16T00:00:00Z","timestamp":1773619200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,3,16]]},"DOI":"10.1109\/percomworkshops68308.2026.11585310","type":"proceedings-article","created":{"date-parts":[[2026,7,2]],"date-time":"2026-07-02T19:42:40Z","timestamp":1783021360000},"page":"93-94","source":"Crossref","is-referenced-by-count":0,"title":["Efficient and Privacy-Preserving Large Language Model Inference at the Edge"],"prefix":"10.1109","author":[{"given":"Vilhelm","family":"Toivonen","sequence":"first","affiliation":[{"name":"University of Helsinki,Department of Computer Science"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Gptq: Accurate post-training quantization for generative pre-trained transformers","author":"Frantar","year":"2023"},{"key":"ref2","article-title":"Awq: Activation-aware weight quantization for on-device llms","author":"Lin","year":"2023"},{"key":"ref3","first-page":"19 274","article-title":"Fast inference from transformers via speculative decoding","volume-title":"ICML","author":"Leviathan"},{"key":"ref4","first-page":"1273","article-title":"Communication-efficient learning of deep networks from decentralized data","volume-title":"Artificial intelligence and statistics","author":"McMahan","year":"2017"},{"key":"ref5","article-title":"Split learning for health: Distributed deep learning without sharing raw patient data","author":"Vepakomma","year":"2018"}],"event":{"name":"2026 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops)","location":"Pisa, Italy","start":{"date-parts":[[2026,3,16]]},"end":{"date-parts":[[2026,3,20]]}},"container-title":["2026 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11585185\/11585186\/11585310.pdf?arnumber=11585310","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,3]],"date-time":"2026-07-03T05:28:00Z","timestamp":1783056480000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11585310\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,16]]},"references-count":5,"URL":"https:\/\/doi.org\/10.1109\/percomworkshops68308.2026.11585310","relation":{},"subject":[],"published":{"date-parts":[[2026,3,16]]}}}