{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,9,11]],"date-time":"2025-09-11T17:54:10Z","timestamp":1757613250068,"version":"3.44.0"},"reference-count":24,"publisher":"IEEE","license":[{"start":{"date-parts":[[2019,10,1]],"date-time":"2019-10-01T00:00:00Z","timestamp":1569888000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2019,10,1]],"date-time":"2019-10-01T00:00:00Z","timestamp":1569888000000},"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":[[2019,10]]},"DOI":"10.1109\/ipccc47392.2019.8958770","type":"proceedings-article","created":{"date-parts":[[2020,1,16]],"date-time":"2020-01-16T21:04:32Z","timestamp":1579208672000},"page":"1-8","source":"Crossref","is-referenced-by-count":5,"title":["Jily: Cost-Aware AutoScaling of Heterogeneous GPU for DNN Inference in Public Cloud"],"prefix":"10.1109","author":[{"given":"Zhaoxing","family":"Wang","sequence":"first","affiliation":[{"name":"University of Chinese Academy of Sciences,School of Cyber Security,Beijing,China,100190"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xuehai","family":"Tang","sequence":"additional","affiliation":[{"name":"Institute of Information Engineering,Chinese Academy of Sciences,Beijing,China,100190"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Qiuyang","family":"Liu","sequence":"additional","affiliation":[{"name":"University of Chinese Academy of Sciences,School of Cyber Security,Beijing,China,100190"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Jizhong","family":"Han","sequence":"additional","affiliation":[{"name":"Institute of Information Engineering,Chinese Academy of Sciences,Beijing,China,100190"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2015.2476459"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1145\/3127479.3132017"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA.2017.52"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1145\/3132847.3133045"},{"journal-title":"AWS Accelerator","year":"2019","key":"ref14"},{"journal-title":"Alibaba Elastic GPU","year":"2019","key":"ref15"},{"key":"ref16","first-page":"1387","article-title":"TFX:A tensorflow-based production-scale machine learning platform","year":"0","journal-title":"Proc 23rd ACM SIGKDD Int Conf Knowl Discov Data Min -KDD"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1002\/cpe.2864"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1145\/3288599.3288634"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-015-0816-y"},{"key":"ref4","first-page":"117","article-title":"Morpheus: Towards automated slos for enterprise clusters","author":"jyothi","year":"0","journal-title":"Proceedings of the 12th USENIX Conference on Operating Systems Design and Implementation OSDI’16"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1145\/3078597.3078604"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1145\/2382553.2382556"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/3135974.3135993"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1016\/j.future.2012.05.018"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1145\/3267809.3267819"},{"journal-title":"TensorFlow Serving","year":"2019","key":"ref2"},{"key":"ref1","first-page":"613","article-title":"Clipper: A low-latency online prediction serving system","author":"crankshaw","year":"0","journal-title":"14th USENIX Symposium on Networked Systems Design and Implementation (NSDI 17)"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1145\/1342171.1342172"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2015.7298965"},{"key":"ref21","article-title":"InferLine: ML Inference Pipeline Composition Framework[J]","author":"crankshaw","year":"2019","journal-title":"arXiv preprint arXiv 1812 01776"},{"journal-title":"The HTK Book Version 3 4","year":"2006","author":"young","key":"ref24"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1145\/2766462.2767708"}],"event":{"name":"2019 IEEE 38th International Performance Computing and Communications Conference (IPCCC)","start":{"date-parts":[[2019,10,29]]},"location":"London, UK","end":{"date-parts":[[2019,10,31]]}},"container-title":["2019 IEEE 38th International Performance Computing and Communications Conference (IPCCC)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/8955479\/8958711\/08958770.pdf?arnumber=8958770","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,9,4]],"date-time":"2025-09-04T18:21:17Z","timestamp":1757010077000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/8958770\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2019,10]]},"references-count":24,"URL":"https:\/\/doi.org\/10.1109\/ipccc47392.2019.8958770","relation":{},"subject":[],"published":{"date-parts":[[2019,10]]}}}