{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,21]],"date-time":"2026-04-21T14:54:19Z","timestamp":1776783259961,"version":"3.51.2"},"reference-count":34,"publisher":"MDPI AG","issue":"10","license":[{"start":{"date-parts":[[2021,10,19]],"date-time":"2021-10-19T00:00:00Z","timestamp":1634601600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100004586","name":"Funda\u00e7\u00e3o Carlos Chagas Filho de Amparo \u00e0 Pesquisa do Estado do Rio de Janeiro","doi-asserted-by":"publisher","award":["E-26\/203.211\/2017"],"award-info":[{"award-number":["E-26\/203.211\/2017"]}],"id":[{"id":"10.13039\/501100004586","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100004586","name":"Funda\u00e7\u00e3o Carlos Chagas Filho de Amparo \u00e0 Pesquisa do Estado do Rio de Janeiro","doi-asserted-by":"publisher","award":["E-26\/202.689\/2018"],"award-info":[{"award-number":["E-26\/202.689\/2018"]}],"id":[{"id":"10.13039\/501100004586","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100004586","name":"Funda\u00e7\u00e3o Carlos Chagas Filho de Amparo \u00e0 Pesquisa do Estado do Rio de Janeiro","doi-asserted-by":"publisher","award":["E-26\/211.144\/2019"],"award-info":[{"award-number":["E-26\/211.144\/2019"]}],"id":[{"id":"10.13039\/501100004586","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001807","name":"S\u00e3o Paulo Research Foundation","doi-asserted-by":"publisher","award":["15\/24494-8"],"award-info":[{"award-number":["15\/24494-8"]}],"id":[{"id":"10.13039\/501100001807","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100002322","name":"Coordena\u00e7\u00e3o de Aperfeicoamento de Pessoal de N\u00edvel Superior","doi-asserted-by":"publisher","award":["001"],"award-info":[{"award-number":["001"]}],"id":[{"id":"10.13039\/501100002322","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100017580","name":"National Education and Research Network","doi-asserted-by":"publisher","award":["0000"],"award-info":[{"award-number":["0000"]}],"id":[{"id":"10.13039\/100017580","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100003593","name":"National Council for Scientific and Technological Development","doi-asserted-by":"publisher","award":["0000"],"award-info":[{"award-number":["0000"]}],"id":[{"id":"10.13039\/501100003593","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Information"],"abstract":"<jats:p>In computer vision applications, mobile devices can transfer the inference of Convolutional Neural Networks (CNNs) to the cloud due to their computational restrictions. Nevertheless, besides introducing more network load concerning the cloud, this approach can make unfeasible applications that require low latency. A possible solution is to use CNNs with early exits at the network edge. These CNNs can pre-classify part of the samples in the intermediate layers based on a confidence criterion. Hence, the device sends to the cloud only samples that have not been satisfactorily classified. This work evaluates the performance of these CNNs at the computational edge, considering an object detection application. For this, we employ a MobiletNetV2 with early exits. The experiments show that the early classification can reduce the data load and the inference time without imposing losses to the application performance.<\/jats:p>","DOI":"10.3390\/info12100431","type":"journal-article","created":{"date-parts":[[2021,10,20]],"date-time":"2021-10-20T03:01:13Z","timestamp":1634698873000},"page":"431","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":20,"title":["Towards Edge Computing Using Early-Exit Convolutional Neural Networks"],"prefix":"10.3390","volume":"12","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-6763-7255","authenticated-orcid":false,"given":"Roberto G.","family":"Pacheco","sequence":"first","affiliation":[{"name":"Grupo de Teleinform\u00e1tica e Automa\u00e7\u00e3o (GTA), PEE\/COPPE-DEL\/Poli, Universidade Federal do Rio de Janeiro (UFRJ), Rio de Janeiro 21941-972, Brazil"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8714-9160","authenticated-orcid":false,"given":"Kaylani","family":"Bochie","sequence":"additional","affiliation":[{"name":"Grupo de Teleinform\u00e1tica e Automa\u00e7\u00e3o (GTA), PEE\/COPPE-DEL\/Poli, Universidade Federal do Rio de Janeiro (UFRJ), Rio de Janeiro 21941-972, Brazil"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0407-2127","authenticated-orcid":false,"given":"Mateus S.","family":"Gilbert","sequence":"additional","affiliation":[{"name":"Grupo de Teleinform\u00e1tica e Automa\u00e7\u00e3o (GTA), PEE\/COPPE-DEL\/Poli, Universidade Federal do Rio de Janeiro (UFRJ), Rio de Janeiro 21941-972, Brazil"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6921-7756","authenticated-orcid":false,"given":"Rodrigo S.","family":"Couto","sequence":"additional","affiliation":[{"name":"Grupo de Teleinform\u00e1tica e Automa\u00e7\u00e3o (GTA), PEE\/COPPE-DEL\/Poli, Universidade Federal do Rio de Janeiro (UFRJ), Rio de Janeiro 21941-972, Brazil"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8752-9382","authenticated-orcid":false,"given":"Miguel Elias M.","family":"Campista","sequence":"additional","affiliation":[{"name":"Grupo de Teleinform\u00e1tica e Automa\u00e7\u00e3o (GTA), PEE\/COPPE-DEL\/Poli, Universidade Federal do Rio de Janeiro (UFRJ), Rio de Janeiro 21941-972, Brazil"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1968","published-online":{"date-parts":[[2021,10,19]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1145\/3042064","article-title":"Deep Learning Advances in Computer Vision with 3D Data: A Survey","volume":"50","author":"Ioannidou","year":"2017","journal-title":"ACM Comput. Surv."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"103213","DOI":"10.1016\/j.jnca.2021.103213","article-title":"A Survey on Deep Learning for Challenged Networks: Applications and Trends","volume":"194","author":"Bochie","year":"2021","journal-title":"J. Netw. Comput. Appl."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"712","DOI":"10.1109\/TITS.2019.2962338","article-title":"A Survey of Deep Learning Applications to Autonomous Vehicle Control","volume":"22","author":"Kuutti","year":"2021","journal-title":"IEEE Trans. Intell. Transp. Syst."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"2295","DOI":"10.1109\/JPROC.2017.2761740","article-title":"Efficient Processing of Deep Neural Networks: A Tutorial and Survey","volume":"105","author":"Sze","year":"2017","journal-title":"Proc. IEEE"},{"key":"ref_5","doi-asserted-by":"crossref","unstructured":"Son, Y., Jeong, J., and Lee, Y. (2018). An Adaptive Offloading Method for an IoT-Cloud Converged Virtual Machine System Using a Hybrid Deep Neural Network. Sustainability, 10.","DOI":"10.3390\/su10113955"},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"25716","DOI":"10.1109\/ACCESS.2021.3058021","article-title":"Learning-in-the-Fog (LiFo): Deep Learning Meets Fog Computing for the Minimum-Energy Distributed Early-Exit of Inference in Delay-Critical IoT Realms","volume":"9","author":"Baccarelli","year":"2021","journal-title":"IEEE Access"},{"key":"ref_7","doi-asserted-by":"crossref","unstructured":"Koci\u0107, J., Jovi\u010di\u0107, N., and Drndarevi\u0107, V. (2019). An End-to-End Deep Neural Network for Autonomous Driving Designed for Embedded Automotive Platforms. Sensors, 19.","DOI":"10.3390\/s19092064"},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"1236","DOI":"10.1093\/bib\/bbx044","article-title":"Deep learning for healthcare: Review, opportunities and challenges","volume":"19","author":"Miotto","year":"2017","journal-title":"Brief. Bioinform."},{"key":"ref_9","doi-asserted-by":"crossref","unstructured":"Teerapittayanon, S., McDanel, B., and Kung, H.T. (2016, January 4\u20138). BranchyNet: Fast inference via early exiting from deep neural networks. Proceedings of the 2016 23rd International Conference on Pattern Recognition (ICPR), Cancun, Mexico.","DOI":"10.1109\/ICPR.2016.7900006"},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"954","DOI":"10.1007\/s12559-020-09734-4","article-title":"Why Should We Add Early Exits to Neural Networks?","volume":"12","author":"Scardapane","year":"2020","journal-title":"Cogn. Comput."},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"Laskaridis, S., Venieris, S.I., Almeida, M., Leontiadis, I., and Lane, N.D. (2020, January 21\u201325). SPINN: Synergistic Progressive Inference of Neural Networks over Device and Cloud. Proceedings of the 26th Annual International Conference on Mobile Computing and Networking (MobiCom\u201920), London, UK.","DOI":"10.1145\/3372224.3419194"},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Sandler, M., Howard, A.G., Zhu, M., Zhmoginov, A., and Chen, L. (2018). Inverted Residuals and Linear Bottlenecks: Mobile Networks for Classification, Detection and Segmentation. arXiv.","DOI":"10.1109\/CVPR.2018.00474"},{"key":"ref_13","unstructured":"Griffin, G., Holub, A., and Perona, P. (2007). Caltech-256 Object Category Dataset, California Institute of Technology. CalTech Report."},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"28","DOI":"10.1109\/MCOM.2012.6178831","article-title":"Challenges of QoE management for cloud applications","volume":"50","author":"Hobfeld","year":"2012","journal-title":"IEEE Commun. Mag."},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"59","DOI":"10.1016\/j.future.2016.09.015","article-title":"Mobile Edge Computing: Opportunities, solutions, and challenges","volume":"70","author":"Ahmed","year":"2017","journal-title":"Future Gener. Comput. Syst."},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Ngo, D., Lee, S., Ngo, T.M., Lee, G.D., and Kang, B. (2021). Visibility Restoration: A Systematic Review and Meta-Analysis. Sensors, 21.","DOI":"10.3390\/s21082625"},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"292","DOI":"10.1109\/JETCAS.2019.2910232","article-title":"Eyeriss v2: A flexible accelerator for emerging deep neural networks on mobile devices","volume":"9","author":"Chen","year":"2019","journal-title":"IEEE J. Emerg. Sel. Top. Circuits Syst."},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"1059","DOI":"10.1007\/s00034-019-01226-7","article-title":"A hardware acceleration platform for AI-based inference at the edge","volume":"39","author":"Karras","year":"2020","journal-title":"Circuits Syst. Signal Process."},{"key":"ref_19","doi-asserted-by":"crossref","first-page":"615","DOI":"10.1145\/3093337.3037698","article-title":"Neurosurgeon: Collaborative intelligence between the cloud and mobile edge","volume":"45","author":"Kang","year":"2017","journal-title":"ACM Comput. Archit. News (SIGARCH)"},{"key":"ref_20","unstructured":"Hu, C., Bao, W., Wang, D., and Liu, F. (May, January 29). Dynamic Adaptive DNN Surgery for Inference Acceleration on the Edge. Proceedings of the IEEE Conference on Computer Communications (INFOCOM), Paris, France."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Pacheco, R.G., and Couto, R.S. (2020, January 7\u201310). Inference Time Optimization Using BranchyNet Partitioning. Proceedings of the 2020 IEEE Symposium on Computers and Communications (ISCC), Rennes, France.","DOI":"10.1109\/ISCC50000.2020.9219647"},{"key":"ref_22","first-page":"1097","article-title":"Imagenet classification with deep convolutional neural networks","volume":"25","author":"Krizhevsky","year":"2012","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref_23","doi-asserted-by":"crossref","unstructured":"Pacheco, R.G., Oliveira, F.D.V.R., and Couto, R.S. (2021, January 7\u201311). Early-exit deep neural networks for distorted images: Providing an efficient edge offloading. Proceedings of the IEEE Global Communications Conference (GLOBECOM), Madrid, Spain. Accepted for publication.","DOI":"10.1109\/GLOBECOM46510.2021.9685469"},{"key":"ref_24","doi-asserted-by":"crossref","first-page":"107346","DOI":"10.1016\/j.patcog.2020.107346","article-title":"Efficient adaptive inference for deep convolutional neural networks using hierarchical early exits","volume":"105","author":"Passalis","year":"2020","journal-title":"Pattern Recognit."},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"4665","DOI":"10.1109\/TII.2018.2842821","article-title":"Deep Learning for Smart Industry: Efficient Manufacture Inspection System With Fog Computing","volume":"14","author":"Li","year":"2018","journal-title":"IEEE Trans. Ind. Inform."},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"30","DOI":"10.1109\/MC.2017.9","article-title":"The emergence of edge computing","volume":"50","author":"Satyanarayanan","year":"2017","journal-title":"Computer"},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"64","DOI":"10.1109\/MCOM.2015.7355568","article-title":"Enhancements of V2X communication in support of cooperative autonomous driving","volume":"53","author":"Hobert","year":"2015","journal-title":"IEEE Commun. Mag."},{"key":"ref_28","doi-asserted-by":"crossref","unstructured":"Chen, Z., Hu, W., Wang, J., Zhao, S., Amos, B., Wu, G., Ha, K., Elgazzar, K., Pillai, P., and Klatzky, R. (2017, January 12\u201314). An empirical study of latency in an emerging class of edge computing applications for wearable cognitive assistance. Proceedings of the Second ACM\/IEEE Symposium on Edge Computing, San Jose, CA, USA.","DOI":"10.1145\/3132211.3134458"},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"436","DOI":"10.1038\/nature14539","article-title":"Deep learning","volume":"521","author":"LeCun","year":"2015","journal-title":"Nature"},{"key":"ref_30","first-page":"1995","article-title":"Convolutional networks for images, speech, and time series","volume":"3361","author":"LeCun","year":"1995","journal-title":"Handb. Brain Theory Neural Netw."},{"key":"ref_31","doi-asserted-by":"crossref","unstructured":"Pacheco, R.G., Couto, R.S., and Simeone, O. (2021, January 14\u201323). Calibration-Aided Edge Inference Offloading via Adaptive Model Partitioning of Deep Neural Networks. Proceedings of the ICC 2021\u2014IEEE International Conference on Communications, Montreal, QC, USA.","DOI":"10.1109\/ICC42927.2021.9500760"},{"key":"ref_32","unstructured":"Guo, C., Pleiss, G., Sun, Y., and Weinberger, K.Q. (2017, January 6\u201311). On calibration of modern neural networks. Proceedings of the 34th International Conference on Machine Learning, Sydney, Australia."},{"key":"ref_33","unstructured":"Howard, A.G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., and Adam, H. (2017). MobileNets: Efficient convolutional neural networks for mobile vision applications. arXiv."},{"key":"ref_34","unstructured":"Pacheco, R.G., Bochie, K., Gilbert, M.S., Couto, R.S., and Campista, M.E.M. (2021, October 08). Available online: https:\/\/github.com\/pachecobeto95\/early_exit_dnn_analysis."}],"container-title":["Information"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2078-2489\/12\/10\/431\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,11]],"date-time":"2025-10-11T07:17:44Z","timestamp":1760167064000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2078-2489\/12\/10\/431"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,10,19]]},"references-count":34,"journal-issue":{"issue":"10","published-online":{"date-parts":[[2021,10]]}},"alternative-id":["info12100431"],"URL":"https:\/\/doi.org\/10.3390\/info12100431","relation":{},"ISSN":["2078-2489"],"issn-type":[{"value":"2078-2489","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,10,19]]}}}