A low-cost implementation method on deep neural network using stochastic computing

Ya Dong; Xingzhong Xiong; Tianyu Li; Lin Zhang; Jienan Chen

doi:10.1117/12.2622719

1 December 2021 A low-cost implementation method on deep neural network using stochastic computing

Ya Dong, Xingzhong Xiong, Tianyu Li, Lin Zhang, Jienan Chen

Proceedings Volume 12079, Second IYSF Academic Symposium on Artificial Intelligence and Computer Engineering; 120792B (2021) https://doi.org/10.1117/12.2622719
Event: 2nd IYSF Academic Symposium on Artificial Intelligence and Computer Engineering, 2021, Xi'an, China

Abstract

The deep neural network (DNN) as the computing core of the multiplication operation consumes a lot of hardware resources, which is not conducive to the development of DNN on a hardware platform with limited resources. In order to solve the above problems, this paper proposes an acceleration method of basing on a deep neural network (DNN) to accelerate multiplication operations on resource-constrained hardware platforms. This method can support sparse calculations to optimize calculation delays. At the same time, inspired by neurosynaptic plasticity and stochastic computing (SC), an acceleration method using simple logic gates for reasoning tasks is proposed. Experimental results show that the hardware resource consumption of the proposed acceleration method is 1.4 times lower than that of the traditional method solution in DNN.

Citation Download Citation

Ya Dong, Xingzhong Xiong, Tianyu Li, Lin Zhang, and Jienan Chen "A low-cost implementation method on deep neural network using stochastic computing", Proc. SPIE 12079, Second IYSF Academic Symposium on Artificial Intelligence and Computer Engineering, 120792B (1 December 2021); https://doi.org/10.1117/12.2622719

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available