Open Access Open Access  Restricted Access Subscription Access

FPGA Based Accelerators of Deep Learning Networks for Learning and Classification


Affiliations
1 Manipal University Jaipur, Dehmi Kalan, Jaipur-Ajmer Express Highway, Jaipur, Rajasthan - 303 007, India
2 Researcher with Manipal Academy of Data Science, MAHE-Bangalore, Karnatak, India

   Subscribe/Renew Journal


A recent trend is to execute computationally intensive algorithms (or work flows) for business analytics using cloud environments which provide machine learning hardware support in the form of GPUs and TPUs. Businesses obtain their data at the sensor level and then perform algorithmic operations on the data via these cloud services. As a result, there can be high input/output data latency which tends to slow down productivity. This thesis work will explore the topic of executing computationally complex algorithms, such as the Convolutional Neural Network (CNN), Recurrent Neural Network(RNN) and Spike Neural Network (SNN), at the sensor level through the use of FPGAs (Field Programmable Gate Arrays) as an alternative to cloud-bound GPU and TPU services.

Keywords

Deep Learning Networks, FPGA.
User
Subscription Login to verify subscription
Notifications
Font Size

  • A.Shawahna, S. M. Sait, and A. El-Maleh, “FPGA-based accelerators of deep learning networks for learning and classification: A review ,” IEEE Access, vol. 7, pp. 7823–7859, 2019. doi: 10.1109/ACCESS.2018.2890150
  • K. Kara, D. Alistarh, G. Alonso, O. Mutlu, and C. Zhang, "FPGA-accelerated dense linear machine learning: A precision-convergence trade-off," In 2017 IEEE 25th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM), Napa ,CA, pp.160–167 , 2017 .doi:10.1109/FCCM.2017.39
  • K. Abdelouahab, M. Pelcat, J. Serot, and F . Berry, "Accelerating CNN Infer ence on FPGA: A Survey ," [Online]. A vailable: https://arxiv .org/abs/1806.01683
  • N. M. Nawi,A. Khan,M.Z.Rehman,H. Chiroma, and T . Herawan, "W eight optimization in Recurrent Neural Networks with Hybrid Metaheuristic Cuckoo Search Techniques for data classification,” Hindawi., Doi:https://doi.org/10.1155/2015/868375
  • A. X. M. Chang, B. Martini, and E. Culurciello, "Recurrent neural networks hardware implementation of FPGA," [Online]. Available:https://arxiv .org/abs/1511.05552
  • J. C. Ferrei ra and J. Fonseca, "An FPGA implementation of a long short-term memory neural network, "In International Conference on ReCon Figurable Computing and FPGAs (ReConFig), Cancun, 2016, pp.1–8.doi:10.1109/ReConFig.2016.7857151
  • D. Honegger, H. Oleynikova and M. Pollefeys, "Real-time and low latency embedded computer vision hardware based on a combination of FPGA and mobile CPU," In 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems, Chicago, IL, pp. 4930 – 4935, 2014. doi:10.1109/IROS.2014.6943263.
  • A. N. Ide and J. H. Saito, "FPGA Implementation of Neocognitrons," In A. R. Omondi, J. C. Rajapakse (eds) FPGA Implementation of Neural Networks, Springer ,Boston, MA.pp-197–224 ,2006. Doi:https://doi.org/10.1007/0-387-28487-7_7
  • K. Kara, D. Alistarh, G. Alonso, O. Mutlu and C. Zhang, "FPGA-accelerated dense l inear machine learning: a precision-convergence trade-off," In IEEE 25th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM), Napa,CA, pp.160–167,2017.IEEE.doi :10.1109/FCCM.2017.39
  • R.Keim, "What is an FPGA? An introduction to programmable logic, " 2018. [Online]. Available: https://www.allaboutcircuits.com/technical -articles/what-is-an-fpga-introduction-to-programmable-logic-fpga-vs-microcontroller/
  • M. Nazemi, S. Nazarian and M. Pedram, "High-performance FPGA implementation of equivariant adaptive separation via independence algorithm for Independent Component Analysis," In 2017 IEEE 28th International Conference on Application- specific Systems, Architectures and Processors (ASAP), Seattle, W A, pp. 25–28, 2017. doi:10.1109/ASAP .2017.7995255
  • K. L. Rice, M. A. Bhuiyan, T . M. T aha, C. N. Vutsinas, and M. C. Smith, "FPGA Implementation of Izhikevich spiking neural networks for character recognition," In 2009 International Conference on Reconfigurable Computing and FPGAs, Quintana Roo, pp. 451 – 456, doi: 10.1109/ReConFig.2009.77
  • J. P . Singh, "Designing an FPGA sythesizable computer vision algorithm to detect the greening of potatoes,"International Journal of Engineering Trends and T echnology , vol. 8, no. 8, pp. 438 – 442, 2014. doi: https://arxiv.org/ct?url=https%3A%2F%2Fdx.doi.or g%2F10.14445%2F22315381%2FIJETT-V8P275&v=9f891483
  • R. Solovyev , A. Kustov, D. Telpukhov , V . Rukhlov and A. Kalinin, "Fixed-point Convolutional Neural Network for real-time video processing in FPGA," In 2019 IEEE Conference of Russian Y oung Researchers in Electrical and Electronic Engineering (EIConRus), Saint Petersburg and Moscow, Russia, pp. 1605 – 1611, 2019. doi:10.1109/EIConRus.2019.8656778
  • D. Soni, "Spiking neural networks, the next generation of machine learning," 2018. [Online]. Avai l abl e: https://towardsdatascience.com/spiking-neural-networks-the-next-generation-of-machine-learning-84e167f4eb2b
  • A. Zuppicich and S. Soltic, "FPGA implementation of an evolving spiking neural network,"In Koppen, M., Kasabov , N., Coghill, G. (eds) Advances in Neuro –Information Processing, International Conference on Neural Iinformation Processing, pp. 1129 – 1136, 2008.

Abstract Views: 8

PDF Views: 0




  • FPGA Based Accelerators of Deep Learning Networks for Learning and Classification

Abstract Views: 8  |  PDF Views: 0

Authors

Subhabaha Pal
Manipal University Jaipur, Dehmi Kalan, Jaipur-Ajmer Express Highway, Jaipur, Rajasthan - 303 007, India
Ravikanth Paturi
Researcher with Manipal Academy of Data Science, MAHE-Bangalore, Karnatak, India

Abstract


A recent trend is to execute computationally intensive algorithms (or work flows) for business analytics using cloud environments which provide machine learning hardware support in the form of GPUs and TPUs. Businesses obtain their data at the sensor level and then perform algorithmic operations on the data via these cloud services. As a result, there can be high input/output data latency which tends to slow down productivity. This thesis work will explore the topic of executing computationally complex algorithms, such as the Convolutional Neural Network (CNN), Recurrent Neural Network(RNN) and Spike Neural Network (SNN), at the sensor level through the use of FPGAs (Field Programmable Gate Arrays) as an alternative to cloud-bound GPU and TPU services.

Keywords


Deep Learning Networks, FPGA.

References





DOI: https://doi.org/10.17010/ijcs%2F2020%2Fv5%2Fi4-5%2F154787