SavedAdd your journal

Journals / Mühendislik Bilimleri ve Araştırmaları Dergisi / 2026 / Cilt: 8 Sayı: 2

FPGA-Based Trainable Neural Network Model Design

FPGA Tabanlı Eğitilebilir Sinirsel Ağ Modeli Tasarımı

Abdelkader Lazzem(SAKARYA UYGULAMALI BİLİMLER ÜNİVERSİTESİ)0000-0003-0136-356X
Halit Öztekin(SAKARYA UYGULAMALI BİLİMLER ÜNİVERSİTESİ)0000-0000-0000-0000
Feyzullah Temurtaş(OSTİM TEKNİK ÜNİVERSİTESİ)1111-1111-0000-0000
Onursal Çetin(TÜRKSAT A.Ş.)1111-1111-1111-1111
Pages
176–185
Publication Date
2026-10-27
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The proposed model is validated using the XOR classification problem, and the experimental results demonstrate successful convergence of the network outputs to the expected values after training.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:This study presents the design and implementation of a trainable artificial neural network (ANN) on a Field-Programmable Gate Array (FPGA).
📊
Key Findings & Data:The proposed architecture consists of artificial neurons with two inputs and one output, employing a 16-bit floating-point representation for arithmetic operations and the Sigmoid activation function.
💡
Conclusion & Impact:The proposed model is validated using the XOR classification problem, and the experimental results demonstrate successful convergence of the network outputs to the expected values after training.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Yapay Sinir Ağı (YSA) · Geri Yayılım · VHDL

Abstract

This study presents the design and implementation of a trainable artificial neural network (ANN) on a Field-Programmable Gate Array (FPGA). The proposed architecture consists of artificial neurons with two inputs and one output, employing a 16-bit floating-point representation for arithmetic operations and the Sigmoid activation function. The network is trained using the backpropagation algorithm to iteratively update the connection weights and improve prediction accuracy. A bottom-up hardware design methodology is adopted, where the ANN is constructed from modular neuron units and their interconnections. The entire architecture is implemented in VHDL using finite state machine (FSM)-based control to achieve a scalable and efficient hardware realization. The proposed model is validated using the XOR classification problem, and the experimental results demonstrate successful convergence of the network outputs to the expected values after training. The findings indicate that the proposed FPGA-based ANN architecture effectively exploits hardware parallelism while providing flexibility and energy efficiency, making it a promising solution for real-time intelligent hardware applications.

Özet

Bu çalışmada, Alan Programlanabilir Kapı Dizisi (FPGA) üzerinde eğitilebilir bir Yapay Sinir Ağı (YSA) modelinin tasarımı ve gerçeklenmesi sunulmaktadır. Önerilen model, iki giriş ve bir çıkışa sahip yapay nöronlardan oluşmakta olup aritmetik işlemlerde 16 bit kayan noktalı sayı gösterimi, aktivasyon fonksiyonu olarak ise Sigmoid fonksiyonu kullanılmaktadır. Ağın eğitimi, bağlantı ağırlıklarının iteratif olarak güncellenmesini sağlayan geri yayılım (Backpropagation) algoritması ile gerçekleştirilmektedir. Donanım tasarımında aşağıdan yukarıya (bottom-up) yaklaşımı benimsenmiş ve YSA mimarisi modüler nöron birimleri ile bu birimler arasındaki bağlantılar temel alınarak oluşturulmuştur. Tasarımın tamamı, ölçeklenebilir ve verimli bir donanım mimarisi elde etmek amacıyla sonlu durum makinesi (FSM) tabanlı kontrol yapısı kullanılarak VHDL dili ile gerçeklenmiştir. Önerilen model, doğrusal olmayan sınıflandırma problemlerinin temel örneklerinden biri olan XOR problemi üzerinde doğrulanmış; elde edilen deneysel sonuçlar, eğitim sürecinin ardından ağ çıktılarının beklenen hedef değerlere başarılı bir şekilde yakınsadığını göstermiştir. Sonuç olarak, önerilen FPGA tabanlı YSA mimarisinin donanımsal paralellikten etkin biçimde yararlanmasının yanı sıra yeniden yapılandırılabilirlik, yüksek işlem performansı ve enerji verimliliği avantajları sunduğu, bu yönleriyle gerçek zamanlı akıllı donanım uygulamaları için uygun ve etkili bir çözüm olduğu değerlendirilmiştir.

🔖

Cite This Article

Lazzem, A., Öztekin, H., Temurtaş, F., Çetin, O. (2026). FPGA-Based Trainable Neural Network Model Design. Mühendislik Bilimleri ve Araştırmaları Dergisi, Cilt: 8 Sayı: 2, 176–185. https://doi.org/10.46387/bjesr.1989621

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (35)

  1. M. Popescu, C. Ravariu, and F. Bulucea, “Neural circuits and their electronic models,” in Proc. 2023 15th Int. Conf. Electronics, Computers and Artificial Intelligence (ECAI), Bucharest, Romania,, pp. 1–4, 2023
  2. T. Riis and J. Kubanek, “Effective ultrasonic stimulation in human peripheral nervous system,” IEEE Trans. Biomed. Eng., vol. 69, no. 1, pp. 15–22, Jan. 2022.
  3. N. M. Safiannikov and P. N. Burenev, “Intelligent measuring system for implementation of the method of assessment of functional state of human central nervous system,” in Proc. 2017 XX IEEE Int. Conf. Soft Computing and Measurements (SCM), St. Petersburg, Russia, pp. 577–579, 2017.
  4. P. Gahiwad, N. Deshmane, S. Karnakar, S. Mali, and R. Pise, “Brain stroke detection using CNN algorithm,” in Proc. 2023 IEEE 8th Int. Conf. for Convergence in Technology (I2CT), Lonavla, India, pp. 1–4, 2023.
  5. S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed. Hoboken, NJ, USA: Pearson, 2021.
  6. C. Aggarwal, “An introduction to neural networks,” in Neural Networks and Deep Learning. Cham, Switzerland: Springer, 2023.
  7. M. Biswas, A. J. Nova, M. K. Mahbub, S. Chaki, S. Ahmed, and M. A. Islam, “Stock market prediction: A survey and evaluation,” in Proc. 2021 Int. Conf. Science & Contemporary Technologies (ICSCT), Dhaka, Bangladesh, pp. 1–6, 2021.
  8. Y. E. Shao, C. H. Wu, B. Y. Ho, and B. S. Hsu, “Integration of multivariate control charts and neural networks to determine the faults of quality characteristic(s) in a multivariate process,” in Proc. 11th Int. Conf. Computer Supported Cooperative Work in Design, Melbourne, VIC, Australia, pp. 1028–1031, 2007.
  9. A. V. Milov, V. S. Tynchenko, and A. V. Murygin, “Neural network modeling to control process of induction soldering,” in Proc. 2019 Int. Conf. Industrial Engineering, Applications and Manufacturing (ICIEAM), Sochi, Russia, pp. 1–5, 2019.
  10. A. V. Kugaevskikh, “COVID-19 screening based on application of neural network classification of exhale spectra,” in Proc. 2022 III Int. Conf. Neural Networks and Neurotechnologies (NeuroNT), St. Petersburg, Russia, pp. 24–27, 2022.
  11. M. S. Chvanova and I. A. Bakalets, “Neural network technology for quality control of the milk packaging process,” in Proc. 2022 6th Scientific School Dynamics of Complex Networks and Their Applications (DCNA), Kaliningrad, Russia, pp. 73–74, 2022.
  12. J. Misra and I. Saha, “Artificial neural networks in hardware: A survey of two decades of progress,” Neurocomputing, vol. 74, no. 1–3, pp. 239–255, 2010.
  13. A. Shrestha, H. Fang, Z. Mei, D. P. Rider, Q. Wu, and Q. Qiu, “A survey on neuromorphic computing: Models and hardware,” IEEE Circuits Syst. Mag., vol. 22, no. 2, pp. 6–35, 2022.
  14. M. Forssell, “Hardware implementation of artificial neural networks,” Information Flow in Networks, course notes, course 18-859E, pp. 1–4, 2014.
  15. C. D. Schuman, T. E. Potok, R. M. Patton, J. D. Birdwell, M. E. Dean, G. S. Rose, and J. S. Plank, “A survey of neuromorphic computing and neural networks in hardware,” arXiv:1705.06963, 2017.
  16. S. S. Lingala, S. Bedekar, P. Tyagi, P. Saha, and P. Shahane, “FPGA-based implementation of neural network,” in Proc. 2022 Int. Conf. Advances in Computing, Communication and Applied Informatics (ACCAI), Chennai, India, pp. 1–5, 2022.
  17. P. D. Deotale and L. E. Dole, “Design of FPGA-based general-purpose neural network,” in Proc. Int. Conf. Information Communication and Embedded Systems (ICICES), Chennai, India, pp. 1–5, 2014.
  18. P. Jawandhiya, “Hardware design for machine learning,” Int. J. Artif. Intell. Appl., vol. 9, no. 1, pp. 63–84, 2018.
  19. J. Liu and D. Liang, “A survey of FPGA-based hardware implementation of ANNs,” in Proc. 2005 Int. Conf. Neural Networks and Brain, Beijing, China, pp. 915–918, 2005.
  20. Z. Li, Y.-J. Huang, and W.-C. Lin, “FPGA implementation of neuron block for artificial neural network,” in Proc. Int. Conf. Electron Devices and Solid-State Circuits (EDSSC), Hsinchu, Taiwan, 2017.
  21. M. Karamimanesh, E. Abiri, M. Shahsavari, K. Hassanli, A. van Schaik, and J. Eshraghian, “Spiking neural networks on FPGA: A survey of methodologies and recent advancements,” Neural Networks, vol. 186, Art. no. 107256, 2025.
  22. J. Ney and N. Wehn, “Achieving high throughput with a trainable neural-network-based equalizer for communications on FPGA,” in Proc. 2024 27th Euromicro Conf. Digital System Design (DSD), Paris, France, pp. 106–113, 2024.
  23. A. Kumar, “FPGA-based hardware accelerators for ANN and DNN,” in Smart Chips for Smart Devices: VLSI Design for Next-Generation IoT Solutions. Hoboken, NJ, USA: Wiley, pp. 31–48, 2026.
  24. D. I. Skrebenkov and D. O. Budanov, “Hardware implementation of neural networks,” in Proc. 2024 Int. Conf. Electrical Engineering and Photonics (EExPolytech), St. Petersburg, Russia, 2024.
  25. K. R. Nichols, M. A. Moussa, and S. M. Areibi, “Feasibility of floating-point arithmetic in FPGA-based artificial neural networks”, 2003.
  26. S. Jeyanthi and M. Subadra, “Implementation of single neuron using various activation functions with FPGA,” in Proc. 2014 IEEE Int. Conf. Advanced Communications, Control and Computing Technologies, pp. 1126–1131, 2014.
  27. S. Haykin, Neural Networks and Learning Machines, 3rd ed. Upper Saddle River, NJ, USA: Pearson Prentice Hall, 2009.
  28. S. K. Abrol and R. Mahajan, “Implementation of single artificial neuron using various activation functions and XOR gate on FPGA chip,” in Proc. 2015 2nd Int. Conf. Advances in Computing and Communication Engineering, pp. 118–123, 2015.
  29. K. Guo, S. Zeng, J. Yu, Y. Wang, and H. Yang, “A survey of FPGA-based neural network inference accelerator,” ACM Trans. Reconfigurable Technol. Syst., vol. 12, no. 1, pp. 1–26, 2019.
  30. P. Micikevicius et al., “Mixed precision training,” in Proc. Int. Conf. Learning Representations (ICLR), 2018.
  31. D. Kalamkar et al., “A study of BFLOAT16 for deep learning training,” arXiv:1905.12322, 2019.
  32. N. Wang, J. Choi, D. Brand, C.-Y. Chen, and K. Gopalakrishnan, “Training deep neural networks with 8-bit floating point numbers,” in Proc. Advances in Neural Information Processing Systems (NeurIPS), 2018.
  33. P. Micikevicius et al., “FP8 formats for deep learning,” arXiv:2209.05433, 2022.
  34. F. Ortega-Zamorano, J. M. Jerez, D. Urda Muñoz, R. M. Luque-Baena, and L. Franco, “Efficient implementation of the backpropagation algorithm in FPGAs and microcontrollers,” IEEE Trans. Neural Netw. Learn. Syst., vol. 27, no. 9, pp. 1840–1850, Sep. 2016.
  35. M. Popescu, C. Ravariu, and F. Bulucea, “Neural circuits and their electronic models,” in Proc. 2023 15th Int. Conf. Electronics, Computers and Artificial Intelligence (ECAI), Bucharest, Romania,, pp. 1–4, 2023