Spiking Neural Network for Object Detection Based on Dual Error

LIU Wei; LI Wenjuan; GAO Jin; LI Liang

doi:10.11999/JEIT221549

Volume 45 Issue 12

Dec. 2023

Turn off MathJax

Article Contents

Article Navigation > Journal of Electronics & Information Technology > 2023 > 45(12): 4469-4476

LIU Wei, LI Wenjuan, GAO Jin, LI Liang. Spiking Neural Network for Object Detection Based on Dual Error[J]. Journal of Electronics & Information Technology, 2023, 45(12): 4469-4476. doi: 10.11999/JEIT221549

Citation:

LIU Wei, LI Wenjuan, GAO Jin, LI Liang. Spiking Neural Network for Object Detection Based on Dual Error[J]. Journal of Electronics & Information Technology, 2023, 45(12): 4469-4476. doi: 10.11999/JEIT221549

Citation:

PDF( 2988 KB)

Spiking Neural Network for Object Detection Based on Dual Error

doi: 10.11999/JEIT221549

LIU Wei^{1
,
,},
LI Wenjuan¹,
GAO Jin¹,
LI Liang²

1.
National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, Beijing 100098, China
2.
Institute of Military Cognition and Brain Sciences, Academy of Military Medical Sciences, Beijing 100850, China

Funds: The National Key Research and Development Program of China (2020AAA0105802, 2020AAA0105800), The National Natural Science Foundation of China (62202469), Beijing Natural Science Foundation (4224091)

Received Date: 2022-12-15
Rev Recd Date: 2023-05-25

Available Online: 2023-06-09

Publish Date: 2023-12-26

Abstract

Abstract

A Spiking Neural Network (SNN) is a low-power neural network that simulates the dynamics of neurons in the brain, providing a feasible solution for deploying object detection tasks in high computational efficiency and low energy consumption environments. Due to the non-differentiable nature of spikes, SNN training is difficult, and a practical solution is to convert pre-trained Artificial Neural Networks (ANNs) into SNNs to improve inference ability. However, the converted SNN often suffers from performance degradation and high latency, which can not meet the high-precision localization requirements for object detection tasks. A dual error is introduced to reduce the loss of conversion performance. To simulate the ANN to SNN conversion error, the causes of errors are analyzed, and a dual error model is built. Further, the dual error model is introduced into the ANN training process so that the errors of the models before and after conversion remain consistent during training and testing, thereby reducing the loss of conversion performance. Finally, the lightweight detection algorithm YOLO is used to verify the effectiveness of the dual error model on the PASCAL VOC and MS COCO datasets.
- Spiking Neural Network (SNN),
- Object detection,
- Conversion error,
- Dual error model

FullText(HTML)

References(29)

References

[1]	DING Xiaohan, ZHANG Xiangyu, MA Ningning, et al. RepVGG: Making VGG-style ConvNets great again[C]. The 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 13728–13737.
[2]	GUO Yiwen, YAO Anbang, and CHEN Yurong. Dynamic network surgery for efficient DNNs[C]. The 30th International Conference on Neural Information Processing Systems, Barcelona, Spain, 2016.
[3]	GOU Jianping, YU Baosheng, MAYBANK S J, et al. Knowledge distillation: A survey[J]. International Journal of Computer Vision, 2021, 129(6): 1789–1819. doi: 10.1007/s11263-021-01453-z
[4]	WANG Lin and YOON K J. Knowledge distillation and student-teacher learning for visual intelligence: A review and new outlooks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 44(6): 3048–3068. doi: 10.1109/TPAMI.2021.3055564
[5]	曲志昱, 李根, 邓志安. 基于知识蒸馏与注意力图的雷达信号识别方法[J]. 电子与信息学报, 2022, 44(9): 3170–3177. doi: 10.11999/JEIT210695 QU Zhiyu, LI Gen, DENG Zhian. Radar signal recognition method based on knowledge distillation and attention map[J]. Journal of Electronics &Information Technology, 2022, 44(9): 3170–3177. doi: 10.11999/JEIT210695
[6]	GONG Yanchao, YANG Kaifang, LIU Ying, et al. Quantization parameter cascading for surveillance video coding considering all inter reference frames[J]. IEEE Transactions on Image Processing, 2021, 30: 5692–5707. doi: 10.1109/TIP.2021.3087413
[7]	JI Mengyu, PENG Gaoliang, LI Sijue, et al. A neural network compression method based on knowledge-distillation and parameter quantization for the bearing fault diagnosis[J]. Applied Soft Computing, 2022, 127: 109331. doi: 10.1016/j.asoc.2022.109331
[8]	AKOPYAN F, SAWADA J, CASSIDY A, et al. TrueNorth: Design and tool flow of a 65 mW 1 million neuron programmable neurosynaptic chip[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2015, 34(10): 1537–1557. doi: 10.1109/TCAD.2015.2474396
[9]	MA Songchen, PEI Jing, ZHANG Weihao, et al. Neuromorphic computing chip with spatiotemporal elasticity for multi-intelligent-tasking robots[J]. Science Robotics, 2022, 7(67): eabk2948. doi: 10.1126/scirobotics.abk2948
[10]	MA De, SHEN Juncheng, GU Zonghua, et al. Darwin: A neuromorphic hardware co-processor based on spiking neural networks[J]. Science China Information Sciences, 2017, 77: 43–51. doi: 10.1016/j.sysarc.2017.01.003
[11]	BU Tong, FANG Wei, DING Jianhao, et al. Optimal ANN-SNN conversion for high-accuracy and ultra-low-latency spiking neural networks[C]. The Tenth International Conference on Learning Representations, Addis Ababa, Ethiopia, 2022.
[12]	CAO Yongqiang, CHEN Yang, and KHOSLA D. Spiking deep convolutional neural networks for energy-efficient object recognition[J]. International Journal of Computer Vision, 2015, 113(1): 54–66. doi: 10.1007/s11263-014-0788-3
[13]	SUN Wei, DAI Liang, ZHANG Xiaorui, et al. RSOD: Real-time small object detection algorithm in UAV-based traffic monitoring[J]. Applied Intelligence, 2022, 52(8): 8448–8463. doi: 10.1007/s10489-021-02893-3
[14]	WU Peirong, LIU Airong, FU Jiyang, et al. Autonomous surface crack identification of concrete structures based on an improved one-stage object detection algorithm[J]. Engineering Structures, 2022, 272: 114962. doi: 10.1016/j.engstruct.2022.114962
[15]	KIM S, PARK S, NA B, et al. Spiking-YOLO: Spiking neural network for energy-efficient object detection[C]. The Thirty-Fourth AAAI Conference on Artificial Intelligence, New York, USA, 2020: 11270–11277.
[16]	SENGUPTA A, YE Yuting, WANG R, et al. Going deeper in spiking neural networks: VGG and residual architectures[J]. Frontiers in Neuroscience, 2019, 13: 95. doi: 10.3389/fnins.2019.00095
[17]	ZHANG Tielin, ZENG Yi, ZHAO Dongcheng, et al. A plasticity-centric approach to train the non-differential spiking neural networks[C]. The Thirty-Second AAAI Conference on Artificial Intelligence, New Orleans, USA, 2018: 77.
[18]	杨静, 吉晓阳, 李少波, 等. 具有正则化约束的脉冲神经网络机器人触觉物体识别方法[J]. 电子与信息学报, 2023, 45(7): 2595–2604. YANG Jing, JI Xiaoyang, LI Shaobo, et al. Spiking neural network robot tactile object recognition method with regularization constraints[J]. Journal of Electronics & Information Technology, 2023, 45(7): 2595–2604.
[19]	MASQUELIER T and THORPE S J. Learning to recognize objects using waves of spikes and spike timing-dependent plasticity[C]. The 2010 International Joint Conference on Neural Networks, Barcelona, Spain, 2010: 1–8.
[20]	BOHTE S M, KOK J N, and LA POUTRÉ J A. SpikeProp: Backpropagation for networks of spiking neurons[C]. The 30th European Symposium on Artificial Neural Networks, Bruges, Belgium, 2000: 419–424.
[21]	HEBB D O. The effect of early and late brain injury upon test scores, and the nature of normal adult intelligence[J]. Proceedings of the American Philosophical Society, 1942, 85(3): 275–292.
[22]	CAPORALE N and DAN Yang. Spike timing-dependent plasticity: A Hebbian learning rule[J]. Annual Review of Neuroscience, 2008, 31: 25–46. doi: 10.1146/annurev.neuro.31.060407.125639
[23]	GERSTNER W. A Framework for Spiking Neuron Models: The Spike Response Model[M]. Handbook of Biological Physics, 2001, 4: 469–516.
[24]	GHAHARI A and ENDERLE J D. A neuron-based time-optimal controller of horizontal saccadic eye movements[J]. International Journal of Neural Systems, 2014, 24(6): 1450017. doi: 10.1142/S0129065714500178
[25]	DIEHL P U, NEIL D, BINAS J, et al. Fast-classifying, high-accuracy spiking deep networks through weight and threshold balancing[C]. 2015 International Joint Conference on Neural Networks, Killarney, Ireland, 2015: 1–8.
[26]	FANG Wei, YU Zhaofei, CHEN Yanqi, et al. Deep residual learning in spiking neural networks[C/OL]. The 35th International Conference on Neural Information Processing Systems, 2021: 21056–21069.
[27]	HAN Bing and ROY K. Deep spiking neural network: Energy efficiency through time based coding[C]. The 16th European Conference on Computer Vision, Glasgow, UK, 2020: 388–404.
[28]	HU Yangfan, TANG Huajin, and PAN Gang. Spiking deep residual networks[J]. IEEE Transactions on Neural Networks and Learning Systems, 2023, 34(8): 5200–5205.
[29]	BU Tong, DING Jianhao, YU Zhaofei, et al. Optimized potential initialization for low-latency spiking neural networks[C]. The AAAI Conference on Artificial Intelligence, Palo Alto, USA, 2022: 11–20.