<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Publishing DTD v1.3 20210610//EN" "JATS-journalpublishing1-3.dtd">
<article article-type="research-article" dtd-version="1.3" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xml:lang="ru"><front><journal-meta><journal-id journal-id-type="publisher-id">pribor</journal-id><journal-title-group><journal-title xml:lang="ru">Известия высших учебных заведений. Приборостроение</journal-title><trans-title-group xml:lang="en"><trans-title>Journal of Instrument Engineering</trans-title></trans-title-group></journal-title-group><issn pub-type="ppub">0021-3454</issn><issn pub-type="epub">2500-0381</issn><publisher><publisher-name>Национальный исследовательский университет ИТМО</publisher-name></publisher></journal-meta><article-meta><article-id pub-id-type="doi">10.17586/0021-3454-2025-68-7-567-575</article-id><article-id custom-type="elpub" pub-id-type="custom">pribor-388</article-id><article-categories><subj-group subj-group-type="heading"><subject>Research Article</subject></subj-group><subj-group subj-group-type="section-heading" xml:lang="ru"><subject>СИСТЕМНЫЙ АНАЛИЗ, УПРАВЛЕНИЕ И ОБРАБОТКА ИНФОРМАЦИИ</subject></subj-group><subj-group subj-group-type="section-heading" xml:lang="en"><subject>SYSTEM ANALYSIS, MANAGEMENT AND INFORMATION PROCESSING</subject></subj-group></article-categories><title-group><article-title>Прямая и обратная задачи оценивания частоты обработки данных в бинарных нейронных сетях</article-title><trans-title-group xml:lang="en"><trans-title>Forward and Inverse Tasks for Inference Rate Estimation in Binary Neural Networks</trans-title></trans-title-group></title-group><contrib-group><contrib contrib-type="author" corresp="yes"><name-alternatives><name name-style="eastern" xml:lang="ru"><surname>Шаккуф</surname><given-names>A.</given-names></name><name name-style="western" xml:lang="en"><surname>Shakkouf</surname><given-names>A.</given-names></name></name-alternatives><bio xml:lang="ru"><p>Али Шаккуф — аспирант; факультет систем управления и робототехники</p><p>Санкт-Петербург</p></bio><bio xml:lang="en"><p>Ali Shakkouf — Post-Graduate Student; Faculty of Control Systems and Robotics</p><p>St. Petersburg</p></bio><email xlink:type="simple">ashakkuf@itmo.ru</email><xref ref-type="aff" rid="aff-1"/></contrib><contrib contrib-type="author" corresp="yes"><name-alternatives><name name-style="eastern" xml:lang="ru"><surname>Громов</surname><given-names>В. С.</given-names></name><name name-style="western" xml:lang="en"><surname>Gromov</surname><given-names>V. S.</given-names></name></name-alternatives><bio xml:lang="ru"><p>Владислав Сергеевич Громов — канд. техн. наук; факультет систем управления и робототехники; доцент</p><p>Санкт-Петербург</p></bio><bio xml:lang="en"><p>Vladislav S. Gromov — PhD; Faculty of Control Systems and Robotics; Associate Professor</p><p>St. Petersburg</p></bio><email xlink:type="simple">gromov@itmo.ru</email><xref ref-type="aff" rid="aff-1"/></contrib></contrib-group><aff-alternatives id="aff-1"><aff xml:lang="ru"><institution>Университет ИТМО</institution><country>Россия</country></aff><aff xml:lang="en"><institution>ITMO University</institution><country>Russian Federation</country></aff></aff-alternatives><pub-date pub-type="collection"><year>2025</year></pub-date><pub-date pub-type="epub"><day>09</day><month>08</month><year>2025</year></pub-date><volume>68</volume><issue>7</issue><fpage>567</fpage><lpage>575</lpage><permissions><copyright-statement>Copyright &amp;#x00A9; Национальный исследовательский университет ИТМО, 2025</copyright-statement><copyright-year>2025</copyright-year><copyright-holder xml:lang="ru">Национальный исследовательский университет ИТМО</copyright-holder><copyright-holder xml:lang="en">Национальный исследовательский университет ИТМО</copyright-holder><license xlink:href="https://pribor.ifmo.ru/jour/about/submissions#copyrightNotice" xlink:type="simple"><license-p>https://pribor.ifmo.ru/jour/about/submissions#copyrightNotice</license-p></license></permissions><self-uri xlink:href="https://pribor.ifmo.ru/jour/article/view/388">https://pribor.ifmo.ru/jour/article/view/388</self-uri><abstract><p>Развертывание бинарных нейронных сетей (BNNS) сопряжено со значительными трудностями, особенно при выборе подходящего оборудования для достижения желаемого уровня производительности и точной оценки вычислительных затрат. Для решения этих проблем была введена новая метрика под названием „XNOROP“, предлагающая упрощенный подход к оценке вычислительных затрат BNNs и представляющая метод сжатия двоичных фильтров. Метрика „XNOROP“ используется для определения и решения двух важнейших задач в BNNs; первая задача, называемая „прямой задачей“, включает в себя оценку скорости вывода данной модели при развертывании на конкретном целевом устройстве; вторая задача, называемая „обратная задача“, описывает систематическую процедуру определения набора целевых устройств, способных обеспечить требуемую скорость вывода при развертывании модели. Расширена базовая формула „XNOROP“ с включением в нее компонентов, сопряженных с временем доступа к памяти, что повысило ее применимость в реальных сценариях развертывания.</p></abstract><trans-abstract xml:lang="en"><p>Deploying Binary Neural Networks (BNNs) presents significant challenges, particularly in selecting suitable hardware to achieve desired performance levels and accurately estimating computational costs. To address these issues, a novel metric named "XNOROP" was recently introduced, offering a simplified approach for estimating the computational cost of BNNs and introducing a method for compressing binary filters. This paper leverages " XNOROP" to define and solve two critical tasks in BNNs. The first task, referred to as the "Forward Task", involves estimating the inference rate of a given model M when deployed on a specific target device T. The second task, known as the "Inverse Task", outlines a systematic procedure to identify a set of target devices T capable of achieving a required inference rate M when deploying the model . Additionally, we extend the foundational formula of "XNOROP" and introduce “LXNOROP” which incorporates considerations for memory access time, enhancing its applicability for real-world deployment scenarios.</p></trans-abstract><kwd-group xml:lang="ru"><kwd>бинарные нейронные сети</kwd><kwd>оптимизация BNNs</kwd><kwd>метрики BNNs</kwd><kwd>XNOROP</kwd><kwd>проектирование аппаратного обеспечения</kwd></kwd-group><kwd-group xml:lang="en"><kwd>Binary Neural Networks</kwd><kwd>BNNs Optimization</kwd><kwd>BNNs metrics</kwd><kwd>XNOROP</kwd><kwd>LXNOROP</kwd><kwd>hardware design</kwd></kwd-group></article-meta></front><back><ref-list><title>References</title><ref id="cit1"><label>1</label><citation-alternatives><mixed-citation xml:lang="ru">Tan S., Zhang Z., Cai Y., Ergu D., Wu L., Hu B., Yu P., &amp; Zhao Y. SegStitch: Multidimensional Transformer for Robust and Efficient Medical Imaging Segmentation // arXiv preprint arXiv:2408.00496, 2024.</mixed-citation><mixed-citation xml:lang="en">Tan S., Zhang Z., Cai Y., Ergu D., Wu L., Hu B., Yu P., &amp; Zhao Y. arXiv preprint, arXiv:2408.00496, 2024.</mixed-citation></citation-alternatives></ref><ref id="cit2"><label>2</label><citation-alternatives><mixed-citation xml:lang="ru">Lin X. V., Shrivastava A., Luo L., Iyer S., Lewis M., Gosh G., Zettlemoyer L., &amp; Aghajanyan A. MoMa: Efficient EarlyFusion Pre-training with Mixture of Modality-Aware Experts // arXiv preprint arXiv:2407.21770, 2024.</mixed-citation><mixed-citation xml:lang="en">Lin X.V., Shrivastava A., Luo L., Iyer S., Lewis M., Gosh G., Zettlemoyer L., &amp; Aghajanyan A. arXiv preprint, arXiv:2407.21770, 2024.</mixed-citation></citation-alternatives></ref><ref id="cit3"><label>3</label><citation-alternatives><mixed-citation xml:lang="ru">Cui Z., Yao J., Zeng L., Yang J., Liu W., &amp; Wang X. LKCell: Efficient Cell Nuclei Instance Segmentation with Large Convolution Kernels // arXiv preprint arXiv: 2407.18054, 2024.</mixed-citation><mixed-citation xml:lang="en">Cui Z., Yao J., Zeng L., Yang J., Liu W., &amp; Wang X. arXiv preprint, arXiv: 2407.18054, 2024.</mixed-citation></citation-alternatives></ref><ref id="cit4"><label>4</label><citation-alternatives><mixed-citation xml:lang="ru">Calculate Computational Efficiency of Deep Learning Models with FLOPs and MACs // KDnuggets. 2023. June 19. [Электронный ресурс]: https://www.kdnuggets.com/2023/06/calculate-computational-efficiency-deep-learningmodels-flops-macs.html</mixed-citation><mixed-citation xml:lang="en">Calculate Computational Efficiency of Deep Learning Models with FLOPs and MACs, KDnuggets, June 19, 2023, https://www.kdnuggets.com/2023/06/calculate-computational-efficiency-deep-learning-models-flops-macs.html.</mixed-citation></citation-alternatives></ref><ref id="cit5"><label>5</label><citation-alternatives><mixed-citation xml:lang="ru">Dongarra J. Performance of various computers using standard linear equations software in a Fortran environment // Argonne National Laboratory Report ANL-80-70. Argonne, Illinois. 1979.</mixed-citation><mixed-citation xml:lang="en">Dongarra J. Performance of various computers using standard linear equations software in a Fortran environment, Argonne National Laboratory Report ANL-80-70, Argonne, Illinois, 1979.</mixed-citation></citation-alternatives></ref><ref id="cit6"><label>6</label><citation-alternatives><mixed-citation xml:lang="ru">Courbariaux M., Bengio Y., &amp; David J.-P. BinaryConnect: Training Deep Neural Networks with binary weights during propagations. 2016. [Электронный ресурс]: https://arxiv.org/abs/1511.00363.</mixed-citation><mixed-citation xml:lang="en">Courbariaux M., Bengio Y., &amp; David J.-P. Binary Connect: Training Deep Neural Networks with binary weights during propagations, 2016, https://arxiv.org/abs/1511.00363.</mixed-citation></citation-alternatives></ref><ref id="cit7"><label>7</label><citation-alternatives><mixed-citation xml:lang="ru">Shakkouf A., Sergeevich G. V. XNOROP: a New Metric for Binary Neural Networks Performance Measurement // 36th Conf. of Open Innovations Association (FRUCT), Lappeenranta, Finland, 2024. P. 721–728. DOI: 10.23919/FRUCT64283.2024.10749876.</mixed-citation><mixed-citation xml:lang="en">Shakkouf A. and Sergeevich G.V. 36th Conference of Open Innovations Association (FRUCT), Lappeenranta, Finland, 2024, pp. 721–728, DOI: 10.23919/FRUCT64283.2024.10749876.</mixed-citation></citation-alternatives></ref><ref id="cit8"><label>8</label><citation-alternatives><mixed-citation xml:lang="ru">Wang L., Zhan J., Gao W., Yang K., Jiang Z., Ren R., He X., &amp; Luo C. BOPS, Not FLOPS! A New Metric and Roofline Performance Model For Datacenter Computing // https://arxiv.org/abs/1801.09212, 2019.</mixed-citation><mixed-citation xml:lang="en">Wang L., Zhan J., Gao W., Yang K., Jiang Z., Ren R., He X., &amp; Luo C. BOPS, Not FLOPS! A New Metric and Roofline Performance Model For Datacenter Computing, 2019, https://arxiv.org/abs/1801.09212.</mixed-citation></citation-alternatives></ref><ref id="cit9"><label>9</label><citation-alternatives><mixed-citation xml:lang="ru">Moosmann J., Bonazzi P., Li Y., Bian S., Mayer P., Benini L., &amp; Magno M. Ultra-Efficient On-Device Object Detection on AI-Integrated Smart Glasses with TinyissimoYOLO // https://arxiv.org/abs/2311.01057, 2023.</mixed-citation><mixed-citation xml:lang="en">Moosmann J., Bonazzi P., Li Y., Bian S., Mayer P., Benini L., &amp; Magno M. Ultra-Efficient On-Device Object Detection on AI-Integrated Smart Glasses with TinyissimoYOLO, 2023, https://arxiv.org/abs/2311.01057.</mixed-citation></citation-alternatives></ref><ref id="cit10"><label>10</label><citation-alternatives><mixed-citation xml:lang="ru">Geiger et al. Larq: An Open-Source Library for Training Binarized Neural Networks // Journal of Open-Source Software. 2020. N 5(45). P. 1746. DOI: 10.21105/joss.01746.</mixed-citation><mixed-citation xml:lang="en">Geiger et al. Journal of Open-Source Software, 2020, no. 5(45), pp. 1746, https://doi.org/10.21105/joss.01746.</mixed-citation></citation-alternatives></ref></ref-list><fn-group><fn fn-type="conflict"><p>The authors declare that there are no conflicts of interest present.</p></fn></fn-group></back></article>
