Open AccessFPGA-based convolutional layer implementation
Author(s) -
A. K. Berzin,
E. S. Dergunov
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1680/1/012006
Subject(s) - field programmable gate array , computer science , convolutional neural network , layer (electronics) , embedded system , computer architecture , parallel computing , artificial intelligence , chemistry , organic chemistry
In this paper, several approaches to the implementation of a convolutional layer of a neural network on FPGAs for use in embedded systems are considered, as well as a number of optimizations necessary to speed up the operation of such a layer. At the end of the work, an FPGA-based implementation is proposed that is comparable in performance with the reference CPU-based one.