Increasing of convolutional neural network performance using residue number system
Nikolay Ivanovich Chervyakov, Pavel Alekseyevich Lyakhov, Maria Vasilyevna Valueva · 2017 International Multi-Conference on Engineering, Computer and Information Sciences (SIBIRCON) · 2017
This paper considers the method of pattern recognition based on a convolutional neural network using Sobel filters. Parameters of the convolutional neural network blocks were chosen experimentally by software modeling in MATLAB. We presents the architecture of the convolutional neural network constructed with residue number system for delay minimization. Using of special type of modules allows to accelerate the work of the device by 37,4% as compared to using a binary number system and by 18,5% as compared to using a known residue number system realization.