Application of the residue number system to reduce hardware costs of the convolutional neural network implementation

DOI10.1016/j.matcom.2020.04.031OpenAlexW3023192521MaRDI QIDQ1998097

N. I. Chervyakov, M. V. Valueva, N. N. Nagornov, G. V. Valuev, P. A. Lyakhov

Publication date: 6 March 2021

Published in: Mathematics and Computers in Simulation (Search for Journal in Brave)

Full work available at URL: https://doi.org/10.1016/j.matcom.2020.04.031

zbMATH Keywords

image processingresidue number systemconvolutional neural networksfield-programmable gate array (FPGA)quantization noise

Mathematics Subject Classification ID

Algorithms in computer science (68Wxx) Theory of computing (68Qxx) Computational number theory (11Yxx) Elementary number theory (11Axx)

Related Items

CRTf-Based Reverse Converter for RNS with Low-Cost Modules $$\{{2}^{n},{2}^{n}-1,{2}^{n+1}-1\}$$, A Study on Affine Transformations and a Novel Universal Prediction Theory, Multi-input bio-inspired weights and structure determination neuronet with applications in European central bank publications, Tool path optimization of selective laser sintering processes using deep learning

Uses Software

• Matlab
• ImageNet
• AlexNet

Cites Work

• An approximate sign detection method for residue numbers and its application to RNS division
• Residue-to-binary conversion for general moduli sets based on approximate Chinese remainder theorem
• On Modulo 2^n+1 Adder Design
• Unnamed Item
• Unnamed Item