DORY: Automatic End-to-End Deployment of Real-World DNNs on Low-Cost IoT MCUs
The deployment of Deep Neural Networks (DNNs) on end-nodes at the extreme edge of the Internet-of-Things is a critical enabler to support pervasive Deep Learning-enhanced applications. Low-Cost MCU-based end-nodes have limited on-chip memory and often replace caches with scratchpads, to reduce area...
Uložené v:
| Vydané v: | IEEE transactions on computers Ročník 70; číslo 8; s. 1253 - 1268 |
|---|---|
| Hlavní autori: | , , , , , |
| Médium: | Journal Article |
| Jazyk: | English |
| Vydavateľské údaje: |
New York
IEEE
01.08.2021
The Institute of Electrical and Electronics Engineers, Inc. (IEEE) |
| Predmet: | |
| ISSN: | 0018-9340, 1557-9956 |
| On-line prístup: | Získať plný text |
| Tagy: |
Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
|
| Abstract | The deployment of Deep Neural Networks (DNNs) on end-nodes at the extreme edge of the Internet-of-Things is a critical enabler to support pervasive Deep Learning-enhanced applications. Low-Cost MCU-based end-nodes have limited on-chip memory and often replace caches with scratchpads, to reduce area overheads and increase energy efficiency - requiring explicit DMA-based memory transfers between different levels of the memory hierarchy. Mapping modern DNNs on these systems requires aggressive topology-dependent tiling and double-buffering. In this work, we propose DORY ( Deployment Oriented to memoRY ) - an automatic tool to deploy DNNs on low cost MCUs with typically less than 1MB of on-chip SRAM memory. DORY abstracts tiling as a Constraint Programming (CP) problem: it maximizes L1 memory utilization under the topological constraints imposed by each DNN layer. Then, it generates ANSI C code to orchestrate off- and on-chip transfers and computation phases. Furthermore, to maximize speed, DORY augments the CP formulation with heuristics promoting performance-effective tile sizes. As a case study for DORY, we target GreenWaves Technologies GAP8, one of the most advanced parallel ultra-low power MCU-class devices on the market. On this device, DORY achieves up to 2.5× better MAC/cycle than the GreenWaves proprietary software solution and 18.1× better than the state-of-the-art result on an STM32-H743 MCU on single layers. Using our tool, GAP-8 can perform end-to-end inference of a 1.0-MobileNet-128 network consuming just 63 pJ/MAC on average @ 4.3 fps - 15.4× better than an STM32-H743. We release all our developments - the DORY framework, the optimized backend kernels, and the related heuristics - as open-source software. |
|---|---|
| AbstractList | The deployment of Deep Neural Networks (DNNs) on end-nodes at the extreme edge of the Internet-of-Things is a critical enabler to support pervasive Deep Learning-enhanced applications. Low-Cost MCU-based end-nodes have limited on-chip memory and often replace caches with scratchpads, to reduce area overheads and increase energy efficiency – requiring explicit DMA-based memory transfers between different levels of the memory hierarchy. Mapping modern DNNs on these systems requires aggressive topology-dependent tiling and double-buffering. In this work, we propose DORY ( Deployment Oriented to memoRY ) – an automatic tool to deploy DNNs on low cost MCUs with typically less than 1MB of on-chip SRAM memory. DORY abstracts tiling as a Constraint Programming (CP) problem: it maximizes L1 memory utilization under the topological constraints imposed by each DNN layer. Then, it generates ANSI C code to orchestrate off- and on-chip transfers and computation phases. Furthermore, to maximize speed, DORY augments the CP formulation with heuristics promoting performance-effective tile sizes. As a case study for DORY, we target GreenWaves Technologies GAP8, one of the most advanced parallel ultra-low power MCU-class devices on the market. On this device, DORY achieves up to 2.5× better MAC/cycle than the GreenWaves proprietary software solution and 18.1× better than the state-of-the-art result on an STM32-H743 MCU on single layers. Using our tool, GAP-8 can perform end-to-end inference of a 1.0-MobileNet-128 network consuming just 63 pJ/MAC on average @ 4.3 fps – 15.4× better than an STM32-H743. We release all our developments – the DORY framework, the optimized backend kernels, and the related heuristics – as open-source software. |
| Author | Garofalo, Angelo Burrello, Alessio Rossi, Davide Tagliavini, Giuseppe Bruschi, Nazareno Conti, Francesco |
| Author_xml | – sequence: 1 givenname: Alessio orcidid: 0000-0002-6215-8220 surname: Burrello fullname: Burrello, Alessio email: alessio.burrello@unibo.it organization: Department of Electrical, Electronic and Information Engineering, University of Bologna, Bologna, Italy – sequence: 2 givenname: Angelo orcidid: 0000-0002-7495-6895 surname: Garofalo fullname: Garofalo, Angelo email: angelo.garofalo@unibo.it organization: Department of Electrical, Electronic and Information Engineering, University of Bologna, Bologna, Italy – sequence: 3 givenname: Nazareno surname: Bruschi fullname: Bruschi, Nazareno email: nazareno.bruschi@unibo.it organization: Department of Electrical, Electronic and Information Engineering, University of Bologna, Bologna, Italy – sequence: 4 givenname: Giuseppe orcidid: 0000-0002-9221-4633 surname: Tagliavini fullname: Tagliavini, Giuseppe email: giuseppe.tagliavini@unibo.it organization: Department of Computer Science and Engineering, University of Bologna, Bologna, Italy – sequence: 5 givenname: Davide orcidid: 0000-0002-0651-5393 surname: Rossi fullname: Rossi, Davide email: davide.rossi@unibo.it organization: Department of Electrical, Electronic and Information Engineering, University of Bologna, Bologna, Italy – sequence: 6 givenname: Francesco orcidid: 0000-0002-7924-933X surname: Conti fullname: Conti, Francesco email: f.conti@unibo.it organization: Department of Electrical, Electronic and Information Engineering, University of Bologna, Bologna, Italy |
| BookMark | eNp9kM9LwzAYhoMoOKdnD14CnrN9aZo08SZ1_oBtgmyIp5J2X6DSNbPJkP33dmx48ODpPXzv873wXJDT1rdIyDWHEedgxot8lEDCRwKU0lqckAGXMmPGSHVKBgBcMyNSOCcXIXwCgErADMjs4fXt447eb6Nf21hXdNKuWPSsD_qAm8bv1thG6h19Q9uwd981_WE-D9S3dOq_We5DpC9-QWf5MlySM2ebgFfHHJLl42SRP7Pp69NLfj9llRAQmXMWLAdljVZVCQatTEtZaSx1iZDpFDNbiUyotFoliSy5dmWJbiUzB2hSFENye_i76fzXFkMsPv22a_vJIpGpyaRKRNK3xodW1fkQOnTFpqvXttsVHIq9s2KRF3tnxdFZT8g_RFXHXotvY2fr5h_u5sDViPi7YoTmimvxAx2HeGA |
| CODEN | ITCOB4 |
| CitedBy_id | crossref_primary_10_1109_TCSI_2022_3222862 crossref_primary_10_1016_j_future_2023_07_020 crossref_primary_10_1007_s00607_025_01514_y crossref_primary_10_1109_TETC_2021_3072337 crossref_primary_10_1109_JETCAS_2023_3328864 crossref_primary_10_1145_3487922 crossref_primary_10_1016_j_iot_2024_101263 crossref_primary_10_1109_JIOT_2021_3091643 crossref_primary_10_1145_3625289 crossref_primary_10_3390_mi12080852 crossref_primary_10_1145_3661820 crossref_primary_10_1016_j_future_2023_07_017 crossref_primary_10_1109_TCSII_2022_3205029 crossref_primary_10_1109_TBCAS_2024_3395534 crossref_primary_10_1109_JSSC_2023_3318301 crossref_primary_10_3390_app14093725 crossref_primary_10_1109_ACCESS_2022_3179047 crossref_primary_10_1109_JETCAS_2022_3170152 crossref_primary_10_1109_TCAD_2022_3232070 crossref_primary_10_3390_electronics12030639 crossref_primary_10_1109_ACCESS_2022_3206782 crossref_primary_10_1109_JIOT_2023_3286276 crossref_primary_10_1109_TBCAS_2021_3122017 crossref_primary_10_1109_JETCAS_2021_3126259 crossref_primary_10_1109_TC_2024_3500369 crossref_primary_10_1109_MCAS_2023_3302182 crossref_primary_10_1109_TC_2024_3354033 crossref_primary_10_1109_JIOT_2023_3331654 crossref_primary_10_1109_TC_2022_3177955 crossref_primary_10_1109_TC_2024_3500360 crossref_primary_10_1109_TBCAS_2024_3478798 crossref_primary_10_1109_JETCAS_2021_3121554 crossref_primary_10_1109_TBCAS_2024_3401858 crossref_primary_10_3390_s23156896 crossref_primary_10_1109_JSSC_2021_3114881 crossref_primary_10_1109_TPDS_2023_3283491 crossref_primary_10_1109_TETC_2023_3322033 crossref_primary_10_1016_j_micpro_2024_105107 crossref_primary_10_1145_3571133 crossref_primary_10_1109_TBCAS_2024_3457522 crossref_primary_10_1109_JSEN_2022_3210773 crossref_primary_10_1109_ACCESS_2022_3146413 crossref_primary_10_1109_MDAT_2025_3573686 crossref_primary_10_3390_jlpea15010008 crossref_primary_10_1109_TBCAS_2025_3573027 crossref_primary_10_1109_TC_2023_3329930 crossref_primary_10_1109_TCSI_2024_3365952 crossref_primary_10_1109_ACCESS_2024_3380472 crossref_primary_10_1109_JIOT_2024_3431913 crossref_primary_10_3390_electronics13142822 crossref_primary_10_1371_journal_pone_0299435 crossref_primary_10_1109_TCAD_2024_3443718 crossref_primary_10_1109_TCSI_2024_3359044 |
| Cites_doi | 10.1145/3007787.3001177 10.1109/CICC.2017.7993626 10.21105/joss.01746 10.1145/3358198 10.1109/CVPR.2018.00474 10.1098/rsta.2019.0155 10.1145/3373376.3378514 10.1007/978-3-030-18338-7_19 10.1109/TCSII.2020.2983648 10.1109/JSSC.2019.2912307 10.1109/ISPASS.2019.00042 10.1109/MCSoC.2015.45 10.1007/s11265-015-1070-9 10.1109/CVPRW.2018.00103 10.1109/MM.2020.2985963 10.1109/JIOT.2019.2917066 10.1109/JIOT.2018.2844296 10.1109/ASAP.2018.8445101 10.1016/j.micpro.2013.04.006 10.1109/MCOM.2017.1600587CM 10.1109/TBCAS.2019.2959160 10.1109/ICCD.2013.6657019 10.1109/DATE.2011.5763085 10.1109/ISSCC.2017.7870349 10.1109/TCSI.2017.2698019 10.1109/TVLSI.2017.2654506 10.1109/CODESISSS.2018.8525915 10.1016/j.dcan.2017.10.002 10.1145/2597917.2597922 |
| ContentType | Journal Article |
| Copyright | Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2021 |
| Copyright_xml | – notice: Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2021 |
| DBID | 97E RIA RIE AAYXX CITATION 7SC 7SP 8FD JQ2 L7M L~C L~D |
| DOI | 10.1109/TC.2021.3066883 |
| DatabaseName | IEEE Xplore (IEEE) IEEE All-Society Periodicals Package (ASPP) 1998–Present IEEE Electronic Library (IEL) CrossRef Computer and Information Systems Abstracts Electronics & Communications Abstracts Technology Research Database ProQuest Computer Science Collection Advanced Technologies Database with Aerospace Computer and Information Systems Abstracts Academic Computer and Information Systems Abstracts Professional |
| DatabaseTitle | CrossRef Technology Research Database Computer and Information Systems Abstracts – Academic Electronics & Communications Abstracts ProQuest Computer Science Collection Computer and Information Systems Abstracts Advanced Technologies Database with Aerospace Computer and Information Systems Abstracts Professional |
| DatabaseTitleList | Technology Research Database |
| Database_xml | – sequence: 1 dbid: RIE name: IEEE Electronic Library (IEL) url: https://ieeexplore.ieee.org/ sourceTypes: Publisher |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Engineering Computer Science |
| EISSN | 1557-9956 |
| EndPage | 1268 |
| ExternalDocumentID | 10_1109_TC_2021_3066883 9381618 |
| Genre | orig-research |
| GrantInformation_xml | – fundername: Artificial Intelligence for Digital Industry grantid: 826060 – fundername: WiPLASH grantid: 863337 – fundername: EU Horizon 2020 Research and Innovation projects OPRECOMP grantid: 732631 |
| GroupedDBID | --Z -DZ -~X .DC 0R~ 29I 4.4 5GY 6IK 85S 97E AAJGR AARMG AASAJ AAWTH ABAZT ABQJQ ABVLG ACGFO ACIWK ACNCT AENEX AETEA AGQYO AHBIQ AKJIK AKQYR ALMA_UNASSIGNED_HOLDINGS ASUFR ATWAV BEFXN BFFAM BGNUA BKEBE BPEOZ CS3 DU5 EBS EJD HZ~ IEDLZ IFIPE IPLJI JAVBF LAI M43 MS~ O9- OCL P2P PQQKQ RIA RIE RNS RXW TAE TN5 TWZ UHB UPT XZL YZZ AAYXX ABUFD CITATION 7SC 7SP 8FD JQ2 L7M L~C L~D |
| ID | FETCH-LOGICAL-c330t-ffa0a106a986cb09ea54b5c8eb8be0784e7ac37364cd225b18fbbefd57f0e94e3 |
| IEDL.DBID | RIE |
| ISICitedReferencesCount | 88 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000671513900010&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| ISSN | 0018-9340 |
| IngestDate | Mon Jun 30 05:42:12 EDT 2025 Sat Nov 29 01:35:42 EST 2025 Tue Nov 18 22:22:11 EST 2025 Wed Aug 27 02:26:38 EDT 2025 |
| IsDoiOpenAccess | false |
| IsOpenAccess | true |
| IsPeerReviewed | true |
| IsScholarly | true |
| Issue | 8 |
| Language | English |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-c330t-ffa0a106a986cb09ea54b5c8eb8be0784e7ac37364cd225b18fbbefd57f0e94e3 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 |
| ORCID | 0000-0002-6215-8220 0000-0002-7495-6895 0000-0002-0651-5393 0000-0002-7924-933X 0000-0002-9221-4633 |
| OpenAccessLink | http://hdl.handle.net/11585/846999 |
| PQID | 2549756232 |
| PQPubID | 85452 |
| PageCount | 16 |
| ParticipantIDs | ieee_primary_9381618 crossref_primary_10_1109_TC_2021_3066883 crossref_citationtrail_10_1109_TC_2021_3066883 proquest_journals_2549756232 |
| PublicationCentury | 2000 |
| PublicationDate | 2021-08-01 |
| PublicationDateYYYYMMDD | 2021-08-01 |
| PublicationDate_xml | – month: 08 year: 2021 text: 2021-08-01 day: 01 |
| PublicationDecade | 2020 |
| PublicationPlace | New York |
| PublicationPlace_xml | – name: New York |
| PublicationTitle | IEEE transactions on computers |
| PublicationTitleAbbrev | TC |
| PublicationYear | 2021 |
| Publisher | IEEE The Institute of Electrical and Electronics Engineers, Inc. (IEEE) |
| Publisher_xml | – name: IEEE – name: The Institute of Electrical and Electronics Engineers, Inc. (IEEE) |
| References | ref13 ref34 ref37 ref15 tan (ref32) 2019 ref14 ref31 han (ref41) 2016 ref11 cecconi (ref19) 2017 ref2 ref1 lai (ref36) 2018 garofalo (ref12) 2020 ref17 abadi (ref39) 2015 ref38 pullini (ref45) 2017 ref18 choi (ref35) 2018 conti (ref40) 2020 lin (ref33) 2020 howard (ref30) 2017 (ref16) 2019 ref23 ref26 ref25 ref20 ref42 david (ref21) 2021 ref44 ref43 ref28 ref27 ref29 ivanov (ref24) 2020 ref8 ref7 ref9 ref4 ref3 ref6 ref5 conti (ref10) 2015 (ref22) 2017 |
| References_xml | – year: 2021 ident: ref21 article-title: TensorFlow lite micro: Embedded machine learning on TinyML systems – year: 2018 ident: ref35 article-title: PACT: Parameterized clipping activation for quantized neural networks – ident: ref9 doi: 10.1145/3007787.3001177 – start-page: 683 year: 2015 ident: ref10 article-title: A ultra-low-energy convolution engine for fast brain-inspired vision in multicore clusters publication-title: Proc Des Autom Test Eur Conf Exhib – ident: ref11 doi: 10.1109/CICC.2017.7993626 – ident: ref29 doi: 10.21105/joss.01746 – ident: ref25 doi: 10.1145/3358198 – ident: ref31 doi: 10.1109/CVPR.2018.00474 – ident: ref14 doi: 10.1098/rsta.2019.0155 – ident: ref27 doi: 10.1145/3373376.3378514 – ident: ref7 doi: 10.1007/978-3-030-18338-7_19 – ident: ref34 doi: 10.1109/TCSII.2020.2983648 – year: 2020 ident: ref24 article-title: Data movement is all you need: A case study of transformer networks – year: 2016 ident: ref41 article-title: Deep compression: Compressing deep neural networks with pruning, trained quantization and huffman coding – ident: ref42 doi: 10.1109/JSSC.2019.2912307 – ident: ref28 doi: 10.1109/ISPASS.2019.00042 – ident: ref18 doi: 10.1109/MCSoC.2015.45 – year: 2019 ident: ref16 article-title: Cypress DRAM – ident: ref15 doi: 10.1007/s11265-015-1070-9 – start-page: 1 year: 2017 ident: ref45 article-title: $\mu$?DMA: An autonomous I/O subsystem for IoT end-nodes publication-title: Proc 27th Int Symp Power Timing Model Optim Simul – ident: ref8 doi: 10.1109/CVPRW.2018.00103 – ident: ref26 doi: 10.1109/MM.2020.2985963 – year: 2015 ident: ref39 article-title: Tensorflow lite for microcontrollers – year: 2020 ident: ref33 article-title: MCUNet: Tiny deep learning on IoT devices – ident: ref5 doi: 10.1109/JIOT.2019.2917066 – ident: ref4 doi: 10.1109/JIOT.2018.2844296 – ident: ref23 doi: 10.1109/ASAP.2018.8445101 – start-page: 186 year: 2020 ident: ref12 article-title: XpulpNN: Accelerating quantized neural networks on RISC-V processors through ISA extensions publication-title: Proc Des Autom Test Eur Conf – ident: ref17 doi: 10.1016/j.micpro.2013.04.006 – start-page: 89 year: 2017 ident: ref19 article-title: Optimal tiling strategy for memory bandwidth reduction for CNNs publication-title: Proc Int Conf Adv Concepts Intell Vis Syst – ident: ref2 doi: 10.1109/MCOM.2017.1600587CM – start-page: 6105 year: 2019 ident: ref32 article-title: EfficientNet: Rethinking model scaling for convolutional neural networks – year: 2018 ident: ref36 article-title: CMSIS-NN: Efficient neural network kernels for arm Cortex-M CPUs – ident: ref3 doi: 10.1109/TBCAS.2019.2959160 – year: 2017 ident: ref22 article-title: X-CUBE-AI – year: 2020 ident: ref40 article-title: Technical Report: NEMO DNN quantization for deployment model – ident: ref20 doi: 10.1109/ICCD.2013.6657019 – year: 2017 ident: ref30 article-title: MobileNets: Efficient convolutional neural networks for mobile vision applications – ident: ref43 doi: 10.1109/DATE.2011.5763085 – ident: ref13 doi: 10.1109/ISSCC.2017.7870349 – ident: ref6 doi: 10.1109/TCSI.2017.2698019 – ident: ref38 doi: 10.1109/TVLSI.2017.2654506 – ident: ref37 doi: 10.1109/CODESISSS.2018.8525915 – ident: ref1 doi: 10.1016/j.dcan.2017.10.002 – ident: ref44 doi: 10.1145/2597917.2597922 |
| SSID | ssj0006209 |
| Score | 2.645775 |
| Snippet | The deployment of Deep Neural Networks (DNNs) on end-nodes at the extreme edge of the Internet-of-Things is a critical enabler to support pervasive Deep... |
| SourceID | proquest crossref ieee |
| SourceType | Aggregation Database Enrichment Source Index Database Publisher |
| StartPage | 1253 |
| SubjectTerms | Acceleration Artificial neural networks C (programming language) Chips (memory devices) Computer architecture Deep neural networks DNN acceleration edge computing Internet of Things IoT Low cost Machine learning Memory management Micromechanical devices Nodes Open source software Software Source code Static random access memory System-on-chip Task analysis Tiling Topology |
| Title | DORY: Automatic End-to-End Deployment of Real-World DNNs on Low-Cost IoT MCUs |
| URI | https://ieeexplore.ieee.org/document/9381618 https://www.proquest.com/docview/2549756232 |
| Volume | 70 |
| WOSCitedRecordID | wos000671513900010&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| journalDatabaseRights | – providerCode: PRVIEE databaseName: IEEE Electronic Library (IEL) customDbUrl: eissn: 1557-9956 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0006209 issn: 0018-9340 databaseCode: RIE dateStart: 19680101 isFulltext: true titleUrlDefault: https://ieeexplore.ieee.org/ providerName: IEEE |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3PT9swFH6CaodxGBs_tEKHfNiBw1ziOo1tbigFMQnKhAqCU2Q7LxLSFE8kZf_-bDetmIADp-RgJ5G_vOfv2c_vA_iuucaqTAVVknGachxTLTNFmdAJt54ByCidcHshplN5d6d-rcGP1VkYRIzJZzgMt3Evv3R2HpbKjlTY5WJyHdaFEIuzWiuvmy3TOZg3YJ4mXRkflqijWe7jwBEbenacScn_m4GipMoLPxwnl7PN933WZ_jUkUhyskD9C6xhvQWbS4EG0tnrFmw8qza4DZeTq-v7Y3Iyb12s00pO65K2jvoLmWAQ_g3vIa4i15490phmQybTaUNcTS7cX5q7piU_3Yxc5jfNDtycnc7yc9qpKVDLedLSqtKJ9gGgVjKzJlGox6kZW4lGGvREIUWhLRc8S23pjdwwWRnjcRyLKkGVIt-FXu1q_AoER5l3E55a8VKmZsS0j3MzZlB5bqC1Un0YLke4sF2p8aB48buIIUeiilleBEiKDpI-HK46_FlU2Xi76XZAYNWsG_w-DJYQFp0VNkUIfkUgeKO913vtw8fw7EVC3wB67eMcv8EH-9Q-NI8H8Qf7BxhiyjQ |
| linkProvider | IEEE |
| linkToHtml | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3BbtQwEB2VgkR7oNBSsVDABw4c8DaOncTmVmVbtWI3oCpF5RTZzkRCqpKqydLfx_ZmVyDgwCk52HLk5xm_icfzAN5prrGpRUaVZJwKjgnVMlWUZTri1jEAGaQTvs6zopDX1-rLFnzY3IVBxJB8hlP_Gs7y684u_a-yY-VPuZh8AA8TIWK2uq218bvpOqGDORPmIhoL-bBIHZe5iwRjNnX8OJWS_7YHBVGVPzxx2F7O9v7vw57Ck5FGkpMV7s9gC9t92FtLNJDRYvdh95d6gwewmH2-_PaRnCyHLlRqJadtTYeOugeZoZf-9eOQriGXjj_SkGhDZkXRk64l8-6e5l0_kIuuJIv8qn8OV2enZX5ORz0FajmPBto0OtIuBNRKptZECnUiTGIlGmnQUQWBmbY846mwtTNzw2RjjEMyyZoIlUB-CNtt1-ILIBinzlE4csVrKUzMtIt0U2ZQOXagtVITmK5nuLJjsXGveXFThaAjUlWZVx6SaoRkAu83HW5XdTb-3fTAI7BpNk7-BI7WEFajHfaVD38zT_Hil3_v9RYen5eLeTW_KD69gh0_ziq97wi2h7slvoZH9sfwvb97ExbbT8nJzXs |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=DORY%3A+Automatic+End-to-End+Deployment+of+Real-World+DNNs+on+Low-Cost+IoT+MCUs&rft.jtitle=IEEE+transactions+on+computers&rft.au=Burrello%2C+Alessio&rft.au=Garofalo%2C+Angelo&rft.au=Bruschi%2C+Nazareno&rft.au=Tagliavini%2C+Giuseppe&rft.date=2021-08-01&rft.pub=The+Institute+of+Electrical+and+Electronics+Engineers%2C+Inc.+%28IEEE%29&rft.issn=0018-9340&rft.eissn=1557-9956&rft.volume=70&rft.issue=8&rft.spage=1253&rft_id=info:doi/10.1109%2FTC.2021.3066883&rft.externalDBID=NO_FULL_TEXT |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0018-9340&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0018-9340&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0018-9340&client=summon |