{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,8,2]],"date-time":"2025-08-02T18:44:23Z","timestamp":1754160263748,"version":"3.41.2"},"reference-count":152,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","license":[{"start":{"date-parts":[[2025,1,1]],"date-time":"2025-01-01T00:00:00Z","timestamp":1735689600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/legalcode"}],"funder":[{"name":"Coordena\u00e7\u00e3o de Aperfei\u00e7oamento de Pessoal de N\u00edvel Superior - CAPES, Brazil"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Access"],"published-print":{"date-parts":[[2025]]},"DOI":"10.1109\/access.2025.3590316","type":"journal-article","created":{"date-parts":[[2025,7,17]],"date-time":"2025-07-17T18:00:13Z","timestamp":1752775213000},"page":"126953-126980","source":"Crossref","is-referenced-by-count":0,"title":["Empirical Evaluation on GPU, Overclocking, and LoRA for Deep Learning on Embedded Systems"],"prefix":"10.1109","volume":"13","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-2705-2225","authenticated-orcid":false,"given":"Evandro","family":"Raphaloski","sequence":"first","affiliation":[{"name":"Federal Institute of Education, Science, and Technology of Rio de Janeiro (IFRJ), Niterói, Rio de Janeiro, Brazil"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1746-8414","authenticated-orcid":false,"given":"Mariana","family":"Caravanti De Souza","sequence":"additional","affiliation":[{"name":"Faculty of Computing (FACOM), Federal University of Mato Grosso do Sul (UFMS), Campo Grande, Mato Grosso do Sul, Brazil"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4471-0886","authenticated-orcid":false,"given":"Edson","family":"Takashi Matsubara","sequence":"additional","affiliation":[{"name":"Faculty of Computing (FACOM), Federal University of Mato Grosso do Sul (UFMS), Campo Grande, Mato Grosso do Sul, Brazil"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/isscc19947.2020.9063049"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1145\/3495532"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1007\/s10489-024-05747-w"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1145\/1150402.1150464"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1503.02531"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/72.80236"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.3390\/electronics11071066"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.3390\/a17010048"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/tcom.1980.1094577"},{"key":"ref10","article-title":"Convolutional neural networks using logarithmic data representation","author":"Miyashita","year":"2016","journal-title":"arXiv:1603.01025"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.3233\/JIFS-169304"},{"key":"ref12","article-title":"Incremental network quantization: Towards lossless CNNs with low-precision weights","author":"Zhou","year":"2017","journal-title":"arXiv:1702.03044"},{"key":"ref13","article-title":"DoReFa-Net: Training low bitwidth convolutional neural networks with low bitwidth gradients","author":"Zhou","year":"2018","journal-title":"arXiv:1606.06160"},{"key":"ref14","first-page":"1","article-title":"LoRA: Low-rank adaptation of large language models","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Hu"},{"key":"ref15","first-page":"1","article-title":"QLoRA: Efficient finetuning of quantized LLMs","volume-title":"Proc. 37th Int. Conf. Neural Inf. Process. Syst.","author":"Dettmers"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.3390\/technologies11020040"},{"key":"ref17","first-page":"1","article-title":"Deep compression: Compressing deep neural network with pruning, trained quantization and Huffman coding","volume-title":"Proc. 4th Int. Conf. Learn. Represent.","author":"Han"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1007\/978-981-19-3391-2_2"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/mlise57402.2022.00011"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/tii.2011.2124466"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1016\/j.iot.2024.101153"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-02686-8_65"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.3390\/mi14050897"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1016\/j.procs.2025.03.273"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1145\/3606376.3593530"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/meco58584.2023.10155048"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.1007\/s42514-024-00208-9"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1109\/depend.2009.30"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/BDCloud.2018.00110"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1109\/tcpmt.2021.3106026"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1109\/lifetech53646.2022.9754820"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/isceic63613.2024.10810178"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1145\/3371154"},{"volume-title":"Models and Pre-trained Weights","year":"2025","key":"ref34"},{"key":"ref35","first-page":"1097","article-title":"ImageNet classification with deep convolutional neural networks","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"25","author":"Krizhevsky"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr.2017.243"},{"key":"ref37","first-page":"6105","article-title":"EfficientNet: Rethinking model scaling for convolutional neural networks","volume-title":"Proc. 36th Int. Conf. Mach. Learn. (ICML)","author":"Tan"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr.2015.7298594"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1109\/iccv.2019.00140"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr.2016.90"},{"key":"ref41","article-title":"Billion-scale semi-supervised learning for image classification","author":"Zeki Yalniz","year":"2019","journal-title":"arXiv:1905.00546"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01264-9_8"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.48550\/ARXIV.1602.07360"},{"key":"ref44","first-page":"1","article-title":"Very deep convolutional networks for large-scale image recognition","volume-title":"Proc. Int. Conf. Learn. Represent. (ICLR)","author":"Simonyan"},{"key":"ref45","article-title":"Mobilenets: Efficient convolutional neural networks for mobile vision applications","author":"Howard","year":"2017","journal-title":"arXiv:1704.04861"},{"volume-title":"NVIDIA CUDNN: GPU-accelerated Deep Learning","year":"2025","key":"ref46"},{"volume-title":"NVIDIA Tensorrt: High-performance Deep Learning Inference","year":"2025","key":"ref47"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr.2019.00293"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.01099"},{"key":"ref50","article-title":"MobileViT: Light-weight, general-purpose, and mobile-friendly vision transformer","author":"Mehta","year":"2022","journal-title":"arXiv:2110.02178"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-19803-8_5"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"ref53","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-015-0816-y"},{"volume-title":"Stable Imagenet-1k","year":"2023","author":"Kinakh","key":"ref54"},{"article-title":"Learning multiple layers of features from tiny images","year":"2009","author":"Krizhevsky","key":"ref55"},{"key":"ref56","article-title":"Efficient hyperparameter optimization in deep learning using a variable length genetic algorithm","author":"Xiao","year":"2020","journal-title":"arXiv:2006.12703"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.1016\/j.jksuci.2020.10.004"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-024-65691-0"},{"key":"ref59","doi-asserted-by":"publisher","DOI":"10.1109\/nas.2017.8026843"},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.1145\/3299710.3211336"},{"key":"ref61","doi-asserted-by":"publisher","DOI":"10.1145\/3448290.3448560"},{"key":"ref62","doi-asserted-by":"publisher","DOI":"10.1109\/tla.2022.9667144"},{"key":"ref63","article-title":"Dora: Weight-decomposed low-rank adaptation","author":"Liu","year":"2024","journal-title":"arXiv:2402.09353"},{"key":"ref64","article-title":"Vera: Vector-based random matrix adaptation","author":"Kopiczko","year":"2024","journal-title":"arXiv:2310.11454"},{"key":"ref65","first-page":"1","article-title":"Do deep nets really need to be deep","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Ba"},{"key":"ref66","doi-asserted-by":"publisher","DOI":"10.1109\/iwqos52092.2021.9521267"},{"key":"ref67","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2020.3039714"},{"key":"ref68","first-page":"1","article-title":"Learning both weights and connections for efficient neural network","volume-title":"Proc. Adv. neural Inf. Process. Syst.","volume":"28","author":"Han"},{"key":"ref69","doi-asserted-by":"publisher","DOI":"10.3390\/electronics11060945"},{"key":"ref70","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2017.541"},{"key":"ref71","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.emnlp-main.279"},{"key":"ref72","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2024.3447085"},{"key":"ref73","doi-asserted-by":"publisher","DOI":"10.1109\/access.2024.3411776"},{"key":"ref74","doi-asserted-by":"publisher","DOI":"10.1109\/access.2019.2954546"},{"key":"ref75","doi-asserted-by":"publisher","DOI":"10.1109\/access.2022.3188323"},{"key":"ref76","doi-asserted-by":"publisher","DOI":"10.1109\/indin45523.2021.9557525"},{"key":"ref77","doi-asserted-by":"publisher","DOI":"10.1109\/jiot.2021.3116316"},{"key":"ref78","doi-asserted-by":"publisher","DOI":"10.1109\/COMPSAC54236.2022.00151"},{"key":"ref79","doi-asserted-by":"publisher","DOI":"10.1109\/MeditCom61057.2024.10621155"},{"key":"ref80","doi-asserted-by":"publisher","DOI":"10.1109\/CVPRW53098.2021.00283"},{"key":"ref81","doi-asserted-by":"publisher","DOI":"10.1016\/j.neucom.2020.12.035"},{"key":"ref82","doi-asserted-by":"publisher","DOI":"10.3390\/asi7010010"},{"key":"ref83","doi-asserted-by":"publisher","DOI":"10.1109\/bigcomp60711.2024.00069"},{"key":"ref84","doi-asserted-by":"crossref","first-page":"137439","DOI":"10.1109\/ACCESS.2024.3455251","article-title":"Mixed-precision neural architecture search and dynamic split point selection for split computing","volume":"12","author":"Nagamatsu","year":"2024","journal-title":"IEEE Access"},{"key":"ref85","doi-asserted-by":"publisher","DOI":"10.1007\/s11265-020-01596-1"},{"key":"ref86","doi-asserted-by":"publisher","DOI":"10.1109\/TCOM.1976.1093309"},{"key":"ref87","doi-asserted-by":"publisher","DOI":"10.1016\/0098-3004(93)90090-R"},{"key":"ref88","doi-asserted-by":"publisher","DOI":"10.1137\/07070111X"},{"key":"ref89","article-title":"Convolutional neural networks with low-rank regularization","author":"Tai","year":"2016","journal-title":"arXiv:1511.06067"},{"key":"ref90","first-page":"1269","article-title":"Exploiting linear structure within convolutional networks for efficient evaluation","volume-title":"Proc. 27th Int. Conf. Neural Inf. Process. Syst.","volume":"1","author":"Denton"},{"key":"ref91","doi-asserted-by":"publisher","DOI":"10.5244\/C.28.88"},{"key":"ref92","doi-asserted-by":"publisher","DOI":"10.1109\/access.2024.3368067"},{"key":"ref93","article-title":"Rethinking the value of network pruning","author":"Liu","year":"2018","journal-title":"arXiv:1810.05270"},{"key":"ref94","article-title":"Model compression via distillation and quantization","author":"Polino","year":"2018","journal-title":"arXiv:1802.05668"},{"key":"ref95","first-page":"1","article-title":"Apprentice: Using knowledge distillation techniques to improve low-precision network accuracy","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Mishra"},{"issue":"50","key":"ref96","first-page":"1","article-title":"Resource-efficient neural networks for embedded systems","volume":"25","author":"Roth","year":"2024","journal-title":"J. Mach. Learn. Res."},{"key":"ref97","doi-asserted-by":"publisher","DOI":"10.30812\/matrik.v23i3.3961"},{"key":"ref98","doi-asserted-by":"publisher","DOI":"10.1145\/3662007.3663881"},{"key":"ref99","doi-asserted-by":"publisher","DOI":"10.1109\/iccvw54120.2021.00092"},{"key":"ref100","doi-asserted-by":"publisher","DOI":"10.1016\/j.aej.2021.11.027"},{"key":"ref101","doi-asserted-by":"publisher","DOI":"10.1016\/j.imavis.2022.104613"},{"key":"ref102","article-title":"Training with mixed-precision floating-point assignments","author":"Lee","year":"2023","journal-title":"arXiv:2301.13464"},{"key":"ref103","doi-asserted-by":"publisher","DOI":"10.1109\/isvlsi49217.2020.000-5"},{"key":"ref104","first-page":"1","article-title":"Dynamic processor overclocking for improving performance of power-constrained systems","volume-title":"IBM Res. Rep.","volume":"2005","author":"Rubio","year":"2005"},{"key":"ref105","doi-asserted-by":"publisher","DOI":"10.1109\/hpcc\/smartcity\/dss.2018.00159"},{"key":"ref106","doi-asserted-by":"publisher","DOI":"10.1109\/iccv51070.2023.01088"},{"key":"ref107","doi-asserted-by":"crossref","DOI":"10.1007\/978-1-4842-2878-4","volume-title":"Raspberry Pi Supercomputing and Scientific Programming: MPI4PY, NumPy, and SciPy for Enthusiasts","author":"Pajankar","year":"2017"},{"volume-title":"NVIDIA Jetson Comparison: Nano, TX2 NX, Xavier NX, AGX, Orin\u2013latest Open Tech From Seeed","year":"2025","key":"ref108"},{"key":"ref109","doi-asserted-by":"publisher","DOI":"10.35784\/jcsi.2437"},{"volume-title":"Overclocking Jetson Nano CPU To 2 GHz and GPU To 1 GHz","year":"2022","key":"ref110"},{"issue":"4","key":"ref111","first-page":"258","article-title":"System level analysis for achieving thermal balance and lifetime reliability in reliably overclocked systems","volume":"2","author":"Ramesh","year":"2010","journal-title":"Int. J. Adv. Syst. Meas."},{"key":"ref112","doi-asserted-by":"publisher","DOI":"10.1109\/iciev.2018.8641023"},{"volume-title":"Safe Overclocking of the Raspberry Pi 4 To 2 GHz","year":"2021","key":"ref113"},{"article-title":"Edge-computing deep learning-based computer vision systems","year":"2018","author":"Weill","key":"ref114"},{"key":"ref115","doi-asserted-by":"publisher","DOI":"10.1109\/isca52012.2021.00055"},{"key":"ref116","doi-asserted-by":"publisher","DOI":"10.1007\/978-1-4302-6425-5_14"},{"key":"ref117","doi-asserted-by":"publisher","DOI":"10.1007\/978-981-16-7443-3_12"},{"key":"ref118","doi-asserted-by":"publisher","DOI":"10.1109\/iceca.2018.8474921"},{"key":"ref119","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/w18-5446"},{"volume-title":"Seq2SQL: Generating Structured Queries From Natural Language Using Reinforcement Learning","year":"2018","author":"Zhong","key":"ref120"},{"key":"ref121","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/d19-5409"},{"key":"ref122","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-84186-7_31"},{"key":"ref123","first-page":"1","article-title":"DeBERTa: Decoding-enhanced BERT with disentangled attention","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"He"},{"issue":"8","key":"ref124","first-page":"9","article-title":"Language models are unsupervised multitask learners","volume":"1","author":"Radford","year":"2019","journal-title":"OpenAI Blog"},{"key":"ref125","first-page":"1877","article-title":"Language models are few-shot learners","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"33","author":"Brown"},{"key":"ref126","first-page":"1","article-title":"Measuring the intrinsic dimension of objective landscapes","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Li"},{"key":"ref127","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.acl-long.568"},{"key":"ref128","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01179"},{"key":"ref129","doi-asserted-by":"publisher","DOI":"10.1109\/icdt61202.2024.10489732"},{"key":"ref130","article-title":"Hyperparameter optimization in machine learning","author":"Franceschi","year":"2024","journal-title":"arXiv:2410.22854"},{"article-title":"Hyperparameter optimization: Foundations, algorithms, best practices and open challenges","year":"2021","author":"Bischl","key":"ref131"},{"key":"ref132","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-05318-5_1"},{"volume-title":"Overclocking the Jetson Nano CPU and GPU","year":"2020","key":"ref133"},{"volume-title":"Throttlestop Application","year":"2025","key":"ref134"},{"volume-title":"Msi Afterburner Application","year":"2025","author":"Afterburner","key":"ref135"},{"volume-title":"Heaven Benchmark Application","year":"2025","key":"ref136"},{"volume-title":"Understanding LoRA and Implementing It","year":"2024","author":"Rasineni","key":"ref137"},{"key":"ref138","first-page":"1","article-title":"Compacter: Efficient low-rank hypercomplex adapter layers","volume-title":"Proc. 35th Int. Conf. Neural Inf. Process. Syst.","author":"Mahabadi"},{"volume-title":"LoRA","year":"2023","author":"Tae","key":"ref139"},{"key":"ref140","first-page":"11727","article-title":"Tensor programs IV: Feature learning in infinite-width neural networks","volume-title":"Proc. 38th Int. Conf. Mach. Learn.","author":"Yang"},{"article-title":"A rank stabilization scaling factor for fine-tuning with LoRA","year":"2023","author":"Kalajdzievski","key":"ref141"},{"key":"ref142","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.naacl-long.282"},{"volume-title":"Practical Guide to Fine-tune LLMs With LoRA","year":"2024","author":"Singh","key":"ref143"},{"article-title":"Allora: Adaptive learning rate mitigates LoRA fatal flaws","year":"2024","author":"Huang","key":"ref144"},{"volume-title":"Different Hyper-parameter Between the Paper and the Code?","year":"2024","author":"Lee","key":"ref145"},{"key":"ref146","article-title":"LoRA dropout as a sparsity regularizer for overfitting control","author":"Lin","year":"2024","journal-title":"arXiv:2404.09610"},{"key":"ref147","doi-asserted-by":"publisher","DOI":"10.3390\/jmse11091658"},{"volume-title":"Power Management for Jetson Nano and Jetson TX1 Devices","year":"2025","key":"ref148"},{"key":"ref149","doi-asserted-by":"publisher","DOI":"10.3390\/computers7010003"},{"key":"ref150","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr.2018.00474"},{"key":"ref151","article-title":"Like what you like: Knowledge distill via neuron selectivity transfer","author":"Huang","year":"2017","journal-title":"arXiv:1707.01219"},{"key":"ref152","article-title":"GaLore: Memory-efficient LLM training by gradient low-rank projection","author":"Zhao","year":"2024","journal-title":"arXiv:2403.03507"}],"container-title":["IEEE Access"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/6287639\/10820123\/11083617.pdf?arnumber=11083617","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,7,26]],"date-time":"2025-07-26T06:32:48Z","timestamp":1753511568000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11083617\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025]]},"references-count":152,"URL":"https:\/\/doi.org\/10.1109\/access.2025.3590316","relation":{},"ISSN":["2169-3536"],"issn-type":[{"type":"electronic","value":"2169-3536"}],"subject":[],"published":{"date-parts":[[2025]]}}}