A method for generating program code based on the parameters of a neural network model
https://doi.org/10.46973/0201-727X_2024_3_50
Abstract
The article proposes a method for generating program code based on the parameters of a neural network model. The necessity of developing this method to improve the performance of software and expand the possibilities of integration into systems with limited hardware resources is justified. The process of software synthesis is considered, including the creation of a direct pass code of a neural network based on the parameters of its model. Examples of code implementation for various configurations of neural networks, including networks with one and several outputs, are given. The results show the possibility of reducing time and resource costs when performing calculations in neural networks, which makes it possible to effectively use lowperformance equipment to perform tasks based on neural networks, which reduces equipment costs. The generated code can be easily integrated into various systems, including those with limited hardware resources. The methodology allows you to adapt and scale solutions to specific tasks and requirements, which is especially important for applied tasks in areas such as image and video processing, speech and text recognition, data analysis and forecasting.
About the Authors
M. N. Cheptsov
Donetsk Institute of Railway Transport
Russian Federation
Russian Federation
Cheptsov Mikhail Nikolaevich, Chair «Automation, telemechanics, communications and computer engineering», Doctor of Technical Sciences, Professor
S. D. Sonina
Donetsk Institute of Railway Transport
Russian Federation
Russian Federation
Sonina Svetlana Dmitrievna, Chair «Automation, telemechanics, communications and computer engineering», Senior Lecturer
M. Yu. Yastrymskiy
Donetsk Institute of Railway Transport
Russian Federation
Russian Federation
Yastrimskiy Maxim Yurievich, Chair «Automation, telemechanics, communications and computer engineering», Graduate Student
References
1. Software environments for studying the basics of neural networks / P. Yu. Bogdanov, E. V. Kraeva, S. A. Verevkin [et al.] // Software products and systems. – 2021. – Vol. 34. – No. 1. – P. 145–150. – DOI 10.15827/0236-235X.133.145-150.
2. Stevens, E. PyTorch. Covering deep learning : professional literature / E. Stevens, T. Wiman, L. Antiga. – Saint-Petersburg : Peter, 2022. – 576 p. – (Series "Programmer's Library"). – ISBN 978-5-4461-1945-5.
3. TensorFlow : LargeScale Machine Learning on Heterogeneous Distributed Systems / M. Abadi, A. Agarwal, P. Barham [et al.] // arXiv preprint arXiv 1603.04467, 2016. – DOI 10.48550/arXiv.1603.04467.
4. Wu, J. (2018). A Comparative Measurement Study of Deep Learning as a Service Framework. IEEE Transactions on Services Computing / J. Wu, Z. Zhang, C. Hsieh // arXiv preprint. – 2022. – Vol. 15. – No. 1. – P. 551–566. – DOI 10.1109/TSC.2019.2928551.
5. Haikin, S. Neural networks : a complete course. Translated from English / S. Haikin. – 2nd ed., rev. – Moscow : “I. D. Williams” LLC, 2006. – 1104 p. – ISBN 5-8459-0890-6.
6. Acceleration of the direct passage in the implementation of CNN on a limited computing resource / A. E. Shchelkunov, V. V. Kovalev, I. V. Sidko, N. E. Sergeev // Izvestiya SFedU. Engineering Sciences. – No. 1 (225). – 2022. – P.289–297. – DOI 10.18522/2311-3103-2022-1-289-297.
7. Review of deep learning : concepts, CNN architectures, challenges, applications, future directions / L. Alzubaidi, J. Zhang, A. J. Humaidi [et al.] // J Big Data 8, 53 (2021). – URL : https://doi.org/10.1186/s40537-021-00444-8 (date of access: 04/09/2024).
8. Sarker, I. H. Deep Learning : A Comprehensive Overview on Techniques, Taxonomy, Applications and Research Directions / I. Sarker. – SN COMPUT. SCI. 2, 420 (2021). – URL: https://doi.org/10.1007/s42979-021-00815-1 (date of access: 04/09/2024).
9. Evaluation of pre-training large language models on leadership-class supercomputers / J. Yin, S. Dash, J. Gounley [et al.] // J Supercomput 79, 20747–20768 (2023). – URL : https://doi.org/10.1007/s11227-023-05479-7 (date of access: 04/09/2024).
10. Cheptsov, M. N. Neural network comparator of real numbers / M. N. Cheptsov, S. D. Sonina // Collection of scientific articles by DONIZHT. – Issue 67. – 2022. – P. 11–15. – ISSN 1993-5579.
Review
For citations:
Cheptsov M.N., Sonina S.D., Yastrymskiy M.Yu. A method for generating program code based on the parameters of a neural network model. Vestnik Rostovskogo Gosudarstvennogo Universiteta Putej Soobshcheniya. 2024;(3):50–55. (In Russ.) https://doi.org/10.46973/0201-727X_2024_3_50
Views:
9
JATS XML
Email this article 