Realization of Artificial Neural Networks on FPGA


ERSOY M. , KUMRAL C. D.

The International Conference on Artificial Intelligence and Applied Mathematics in Engineering, Antalya, Turkey, 18 - 20 April 2019, vol.43, pp.418-428 identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 43
  • Doi Number: 10.1007/978-3-030-36178-5_31
  • City: Antalya
  • Country: Turkey
  • Page Numbers: pp.418-428
  • Keywords: Artificial Neural Networks, Field Programmable Gate Array, Very High Speed Integrated Circuit Hardware Description Language, Real time systems

Abstract

Artificial Neural Networks (ANNs) are generally modeled and used as software based. Software models are insufficient in real time applications where ANN output needs to be calculated. ANN has an architecture that can operate in parallel to calculate hidden layers. The fact that ANN has such an architecture makes it potentially fast in calculating certain transactions. However, the speed of these operations in real-time systems depends on the specification of the hardware. Therefore, ANN design has been realized on FPGA which is capable of parallel processing. In this way, the ANN structure was realized in a hardware structure and it was provided to be used on real-time structures.