RT Journal Article
A1 Zhitao Lin
A1 Juncheng Shen
A1 De Ma
A1 Jianyi Meng

PB iet
T1 Quantisation and pooling method for low-inference-latency spiking neural networks
JN Electronics Letters
VO 53
IS 20
SP 1347
OP 1348
AB Spiking neural network (SNN) that converted from conventional deep neural network (DNN) has shown great potential as a solution for fast and efficient recognition. A layer-wise quantisation method based on retraining is proposed to quantise the activation of DNN, which reduces the number of time steps required by converted SNN to achieve minimal accuracy loss. Pooling function is incorporated into convolutional layers to reduce at most 20% of spiking neurons. The converted SNNs achieved 99.15% accuracy on MNIST and 82.9% on CIFAR10 by only seven time steps, and only 10–40% of spikes need to be processed compared with networks using traditional algorithms. The experimental results show that the proposed methods are able to build hardware-friendly SNNs with ultra-low-inference latency.
K1 DNN
K1 spiking neurons
K1 layer-wise quantisation method
K1 SNN
K1 CIFAR10
K1 retraining
K1 MNIST
K1 real-time recognition tasks
K1 pooling method
K1 convolutional layers
K1 deep neural network
K1 pooling function
K1 low-inference-latency spiking neural networks
DO https://doi.org/10.1049/el.2017.2219
UL https://digital-library.theiet.org/;jsessionid=4ujmofp54abj.x-iet-live-01content/journals/10.1049/el.2017.2219
LA English
SN 0013-5194
YR 2017
OL EN