{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,9]],"date-time":"2026-07-09T20:21:54Z","timestamp":1783628514086,"version":"3.55.0"},"reference-count":39,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"name":"National Key Research and Development Program of China","award":["2023YFB2703600"],"award-info":[{"award-number":["2023YFB2703600"]}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62572280"],"award-info":[{"award-number":["62572280"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["U24A20244"],"award-info":[{"award-number":["U24A20244"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Netw."],"published-print":{"date-parts":[[2026]]},"DOI":"10.1109\/ton.2026.3704584","type":"journal-article","created":{"date-parts":[[2026,6,17]],"date-time":"2026-06-17T19:37:55Z","timestamp":1781725075000},"page":"5942-5953","source":"Crossref","is-referenced-by-count":0,"title":["Efficient Mixture-of-Experts Model Inference at the Edge via Adaptive Expert Merging"],"prefix":"10.1109","volume":"34","author":[{"ORCID":"https:\/\/orcid.org\/0009-0006-2346-2031","authenticated-orcid":false,"given":"Ruirui","family":"Zhang","sequence":"first","affiliation":[{"name":"Department of Computer Science, Shandong University, Qingdao, Shandong, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4579-5380","authenticated-orcid":false,"given":"Yifei","family":"Zou","sequence":"additional","affiliation":[{"name":"Department of Computer Science, Shandong University, Qingdao, Shandong, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5303-0700","authenticated-orcid":false,"given":"Peng","family":"Li","sequence":"additional","affiliation":[{"name":"Department of Computer Science, Xi’an Jiaotong University, Xi’an, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Fahao","family":"Chen","sequence":"additional","affiliation":[{"name":"School of Artificial Intelligence, Shandong University, Jinan, Shandong, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yupeng","family":"Li","sequence":"additional","affiliation":[{"name":"Department of Interactive Media, Hong Kong Baptist University, Kowloon Tsai, Hong Kong"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5912-4647","authenticated-orcid":false,"given":"Xiuzhen","family":"Cheng","sequence":"additional","affiliation":[{"name":"Shandong Provincial Key Laboratory of Computing-Network Integration, Qingdao, Shandong, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1989-1750","authenticated-orcid":false,"given":"Falko","family":"Dressler","sequence":"additional","affiliation":[{"name":"School of Electrical Engineering and Computer Science, TU Berlin, Berlin, Germany"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6835-5981","authenticated-orcid":false,"given":"Dongxiao","family":"Yu","sequence":"additional","affiliation":[{"name":"School of Cryptography Science and Engineering, Shandong University, Jinan, Shandong, China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Outrageously large neural networks: The sparsely-gated mixture-of-experts layer","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Shazeer"},{"key":"ref2","article-title":"Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity","author":"Fedus","year":"2021","journal-title":"arXiv:2101.03961"},{"key":"ref3","article-title":"Soft merging of experts with adaptive routing","author":"Muqeeth","year":"2023","journal-title":"arXiv:2306.03745"},{"key":"ref4","article-title":"PopulAtion parameter averaging (PAPA)","author":"Jolicoeur-Martineau","year":"2023","journal-title":"arXiv:2304.03094"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.emnlp-main.907"},{"key":"ref6","article-title":"Learning more generalized experts by merging experts in mixture-of-experts","author":"Park","year":"2024","journal-title":"arXiv:2405.11530"},{"key":"ref7","article-title":"Merge, then compress: Demystify efficient SMoE with hints from its routing policy","author":"Li","year":"2023","journal-title":"arXiv:2310.01334"},{"key":"ref8","first-page":"18015","article-title":"$S^{3}$\n: Increasing GPU utilization during generative inference for higher throughput","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"36","author":"Brooks"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM48880.2022.9796961"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/71.993206"},{"issue":"3","key":"ref11","first-page":"6","article-title":"An improved weighted least connection scheduling algorithm for load balancing in web cluster systems","volume":"5","author":"Singh","year":"2018","journal-title":"Int. Res. J. Eng. Technol."},{"key":"ref12","first-page":"26433","article-title":"QoS-efficient serving of multiple mixture-of-expert LLMs using partial runtime reconfiguration","volume-title":"Proc. 42nd Int. Conf. Mach. Learn.","volume":"267","author":"Imani"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.52202\/079017-2670"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/ICRC64395.2024.10937027"},{"key":"ref15","first-page":"90","article-title":"Choice of plausible alternatives: An evaluation of commonsense causal reasoning","volume-title":"Proc. AAAI Spring Symp. Ser.","author":"Roemmele"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/N18-1023"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D16-1264"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D15-1237"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1145\/3474381"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1080\/00207543.2016.1162341"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1016\/j.hcc.2024.100270"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM52122.2024.10621342"},{"key":"ref23","first-page":"22941","article-title":"On-device training under 256KB memory","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Chen"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1016\/j.hcc.2024.100296"},{"key":"ref25","article-title":"MoE-prism: Disentangling monolithic experts for elastic MoE services via model-system co-designs","author":"Xia","year":"2025","journal-title":"arXiv:2510.19366"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1145\/3767295.3769319"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.1145\/3721146.3721940"},{"key":"ref28","first-page":"269","article-title":"Tutel: Adaptive mixture-of-experts at scale","volume-title":"Proc. Mach. Learn. Syst.","volume":"5","author":"Hwang"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1145\/3718958.3750506"},{"key":"ref30","article-title":"A white paper on neural network quantization","author":"Nagel","year":"2021","journal-title":"arXiv:2106.08295"},{"key":"ref31","article-title":"DynaMo: Runtime switchable quantization for MoE with cross-dataset adaptation","author":"Zheng","year":"2025","journal-title":"arXiv:2503.21135"},{"key":"ref32","article-title":"Task-specific expert pruning for sparse mixture-of-experts","author":"Chen","year":"2022","journal-title":"arXiv:2206.00277"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1007\/978-981-96-4207-6_12"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v38i14.29539"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.3390\/electronics13112077"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1109\/TMC.2025.3546466"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1109\/IWCMC61514.2024.10592339"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/TON.2025.3649068"},{"key":"ref39","article-title":"Hybrid LLM: Cost-efficient and quality-aware query routing","author":"Ding","year":"2024","journal-title":"arXiv:2404.14618"}],"container-title":["IEEE Transactions on Networking"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/10723154\/11317935\/11569965.pdf?arnumber=11569965","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,9]],"date-time":"2026-07-09T19:42:37Z","timestamp":1783626157000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11569965\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":39,"URL":"https:\/\/doi.org\/10.1109\/ton.2026.3704584","relation":{},"ISSN":["2998-4157"],"issn-type":[{"value":"2998-4157","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026]]}}}