Paper Info
Title
HEIF: Highly Efficient Stochastic Computing based Inference Framework for Deep Neural Networks
Abstract
Deep convolutional neural networks (DCNNs) are one of the most promising deep learning techniques and have been recognized as the dominant approach for almost all recognition and detection tasks. The computation of DCNNs is memory intensive due to large feature maps and neuron connections, and the performance highly depends on the capability of hardware resources. With the recent trend of wearable devices and Internet of Things, it becomes desirable to integrate the DCNNs onto embedded and portable devices that require low power and energy consumptions and small hardware footprints. Recently stochastic computing (SC)-DCNN demonstrated that SC as a low-cost substitute to binary-based computing radically simplifies the hardware implementation of arithmetic units and has the potential to satisfy the stringent power requirements in embedded devices. In SC, many arithmetic operations that are resource-consuming in binary designs can be implemented with very simple hardware logic, alleviating the extensive computational complexity. It offers a colossal design space for integration and optimization due to its reduced area and soft error resiliency. In this paper, we present HEIF, a highly efficient SC-based inference framework of the large-scale DCNNs, with broad applications including (but not limited to) <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LeNet-5 and AlexNet</italic> , that achieves high energy efficiency and low area/hardware cost. Compared to SC-DCNN, HEIF features: 1) the first (to the best of our knowledge) SC-based rectified linear unit activation function to catch up with the recent advances in software models and mitigate degradation in application-level accuracy; 2) the redesigned approximate parallel counter and optimized stochastic multiplication using transmission gates and inverse mirror adders; and 3) the new optimization of weight storage using clustering. Most importantly, to achieve maximum energy efficiency while maintaining acceptable accuracy, HEIF considers holistic optimizations on cascade connection of function blocks in DCNN, pipelining technique, and bit-stream length reduction. Experimental results show that in large-scale applications HEIF outperforms previous SC-DCNN by the throughput of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$4.1\times $ </tex-math></inline-formula> , by area efficiency of up to <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$6.5\times $ </tex-math></inline-formula> , and achieves up to <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${5.6\times }$ </tex-math></inline-formula> energy improvement.
Year
DOI
Venue
2019
10.1109/tcad.2018.2852752
IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
Keywords
Field
DocType
Hardware,Feature extraction,Machine learning,Neurons,Optimization,Convolutional neural networks
Pipeline (computing),Rectifier (neural networks),Soft error,Computer science,Efficient energy use,Convolutional neural network,Electronic engineering,Artificial intelligence,Deep learning,Computer engineering,Stochastic computing,Computational complexity theory
Journal
Volume
Issue
ISSN
38
8
0278-0070
Citations 
PageRank 
References 
8
0.50
0
Authors
11
Name
Order
Citations
PageRank
Qinru Qiu11120102.58
Qinru Qiu21120102.58
Ji Li39710.87
Ao Ren49611.53
Ruizhe Cai591.53
Caiwen Ding614226.52
Xuehai Qian732027.71
Jeff Draper829826.31
Bo Yuan926228.64
Jian Tang10109574.34
Yanzhi Wang111082136.11