{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,1]],"date-time":"2026-07-01T18:23:03Z","timestamp":1782930183917,"version":"3.54.5"},"reference-count":60,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"4","license":[{"start":{"date-parts":[[2021,12,1]],"date-time":"2021-12-01T00:00:00Z","timestamp":1638316800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2021,12,1]],"date-time":"2021-12-01T00:00:00Z","timestamp":1638316800000},"content-version":"am","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"}],"funder":[{"DOI":"10.13039\/501100011688","name":"ECSEL Horizon 2020 Project Artificial intelligence for Digital Industry","doi-asserted-by":"publisher","award":["826060"],"award-info":[{"award-number":["826060"]}],"id":[{"id":"10.13039\/501100011688","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100010669","name":"EU Horizon 2020 Project BonsAPPs","doi-asserted-by":"publisher","award":["101015848"],"award-info":[{"award-number":["101015848"]}],"id":[{"id":"10.13039\/100010669","id-type":"DOI","asserted-by":"publisher"}]},{"name":"CINECA through the ISCRA Initiative under Project NAS4NPC"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE J. Emerg. Sel. Topics Circuits Syst."],"published-print":{"date-parts":[[2021,12]]},"DOI":"10.1109\/jetcas.2021.3121554","type":"journal-article","created":{"date-parts":[[2021,10,20]],"date-time":"2021-10-20T18:02:18Z","timestamp":1634752938000},"page":"789-802","source":"Crossref","is-referenced-by-count":90,"title":["A TinyML Platform for On-Device Continual Learning With Quantized Latent Replays"],"prefix":"10.1109","volume":"11","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-9557-559X","authenticated-orcid":false,"given":"Leonardo","family":"Ravaglia","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7458-4019","authenticated-orcid":false,"given":"Manuele","family":"Rusci","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Davide","family":"Nadalini","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8705-0761","authenticated-orcid":false,"given":"Alessandro","family":"Capotondi","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7924-933X","authenticated-orcid":false,"given":"Francesco","family":"Conti","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8068-3806","authenticated-orcid":false,"given":"Luca","family":"Benini","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2021.3066883"},{"key":"ref38","first-page":"578","article-title":"TVM: An automated end-to-end optimizing compiler for deep learning","author":"chen","year":"2018","journal-title":"Proc of USENIX Symp on Operating Systems Design and Implementation (OSDI)"},{"key":"ref33","article-title":"Deep compression: Compressing deep neural networks with pruning, trained quantization and Huffman coding","author":"han","year":"2016","journal-title":"arXiv 1510 00149 [cs]"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/JPROC.2017.2761740"},{"key":"ref31","first-page":"246","article-title":"Envision: A 0.26-to-10 TOPS\/W subword-parallel dynamic-voltage-accuracy-frequency-scalable convolutional neural network processor in 28 nm FDSOI","author":"moons","year":"2017","journal-title":"IEEE Int Solid-State Circuits Conf (ISSCC) Dig Tech Papers"},{"key":"ref30","first-page":"1240","article-title":"Online learned continual compression with adaptive quantization modules","author":"caccia","year":"2020","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref37","article-title":"CMSIS-NN: Efficient neural network kernels for arm cortex-M CPUs","author":"lai","year":"2018","journal-title":"arXiv 1801 06601 [cs]"},{"key":"ref36","first-page":"1","article-title":"512 KiB RAM is enough! live camera face recognition DNN on MCU","author":"zemlyanikin","year":"2019","journal-title":"Proc IEEE\/CVF Int Conf Comput Vis Workshop"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1038\/s41928-018-0054-8"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2016.2616357"},{"key":"ref60","article-title":"Technical report: NEMO DNN quantization for deployment model","author":"conti","year":"2020","journal-title":"arXiv 2004 05930"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.587"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01258-8_15"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/ICRA.2019.8793982"},{"key":"ref2","first-page":"1935","article-title":"Resource-efficient machine learning in 2 kb RAM for the Internet of Things","author":"kumar","year":"2017","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/IROS45743.2020.9341460"},{"key":"ref20","first-page":"60","article-title":"A 1.3 TOPS\/W 32 GOPS fully integrated 10-core SoC for IoT end-nodes with 1.7 $\\mu\\text{W}$\n cognitive wake-up from MRAM-based state-retentive sleep mode","volume":"64","author":"rossi","year":"2021","journal-title":"IEEE Int Solid-State Circuits Conf (ISSCC) Dig Tech Papers"},{"key":"ref22","first-page":"1","article-title":"TinyTL: Reduce memory, not parameters for efficient on-device learning","volume":"33","author":"cai","year":"2020","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/MSPEC.2019.8701189"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1145\/3417313.3429378"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/IJCNN52387.2021.9533927"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1016\/j.neunet.2019.03.010"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1109\/TBCAS.2019.2914476"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2021.3066400"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC19947.2020.9062989"},{"key":"ref59","doi-asserted-by":"publisher","DOI":"10.1109\/CVPRW50498.2020.00131"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1109\/TMSCS.2017.2769046"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.1109\/ISCAS.2018.8351816"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00286"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-021-22364-0"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-019-1424-8"},{"key":"ref53","doi-asserted-by":"publisher","DOI":"10.1109\/MM.2018.112130359"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1016\/j.neunet.2019.09.004"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1611835114"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1145\/3450494"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1109\/TCSII.2020.2983648"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2021.3057446"},{"key":"ref13","article-title":"Online continual learning in image classification: An empirical survey","author":"mai","year":"2021","journal-title":"arXiv 2101 10423"},{"key":"ref14","article-title":"Three scenarios for continual learning","author":"van de ven","year":"2019","journal-title":"arXiv 1904 07734"},{"key":"ref15","article-title":"On tiny episodic memories in continual learning","author":"chaudhry","year":"2019","journal-title":"arXiv 1902 10486"},{"key":"ref16","article-title":"CVPR 2020 continual learning in computer vision competition: Approaches, results, current challenges and future directions","author":"lomonaco","year":"2020","journal-title":"arXiv 2009 09929"},{"key":"ref17","article-title":"Batch-level experience replay with review for continual learning","author":"mai","year":"2020","journal-title":"arXiv 2007 05683"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1109\/SiPS50750.2020.9195220"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/HOTCHIPS.2015.7477325"},{"key":"ref4","article-title":"PACT: Parameterized clipping activation for quantized neural networks","author":"choi","year":"2018","journal-title":"arXiv 1805 06085"},{"key":"ref3","article-title":"Benchmarking tinyML systems: Challenges and direction","author":"banbury","year":"2020","journal-title":"arXiv 2003 04821"},{"key":"ref6","article-title":"TensorFlow lite micro: Embedded machine learning on TinyML systems","author":"david","year":"2020","journal-title":"arXiv 2010 08678"},{"key":"ref5","first-page":"129","article-title":"What is the state of neural network pruning?","volume":"2","author":"blalock","year":"2020","journal-title":"Proc Mach Learn Syst"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.3390\/s21041339"},{"key":"ref7","article-title":"Concrete problems in AI safety","author":"amodei","year":"2016","journal-title":"arXiv 1606 06565"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1109\/DAC18072.2020.9218507"},{"key":"ref9","first-page":"92","article-title":"In-situAI: Towards autonomous and incremental deep learning for IoT systems","author":"song","year":"2018","journal-title":"Proc IEEE Int Symp High Perform Comput Architecture (HPCA)"},{"key":"ref46","first-page":"4900","article-title":"Hybrid 8-bit floating point (HFP8) training and inference for deep neural networks","volume":"32","author":"sun","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref45","article-title":"A study of BFLOAT16 for deep learning training","author":"kalamkar","year":"2019","journal-title":"arXiv 1905 12322"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1145\/2654822.2541967"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2017.7870350"},{"key":"ref42","article-title":"Quantization-guided training for compact tinyML models","author":"ghamari","year":"2021","journal-title":"arXiv 2103 06231"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2017.2774288"},{"key":"ref44","doi-asserted-by":"publisher","DOI":"10.1109\/MM.2019.2928962"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-70353-4_8"}],"container-title":["IEEE Journal on Emerging and Selected Topics in Circuits and Systems"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/5503868\/9647029\/09580920.pdf?arnumber=9580920","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,3,8]],"date-time":"2022-03-08T21:46:51Z","timestamp":1646776011000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/9580920\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,12]]},"references-count":60,"journal-issue":{"issue":"4"},"URL":"https:\/\/doi.org\/10.1109\/jetcas.2021.3121554","relation":{},"ISSN":["2156-3357","2156-3365"],"issn-type":[{"value":"2156-3357","type":"print"},{"value":"2156-3365","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,12]]}}}