Spiking Neural Network (SNN) With Memristor Synapses Having Non-linear Weight Update

被引：62

作者：

Kim, Taeyoon ^{[1
]}

Hu, Suman ^{[1
]}

Kim, Jaewook ^{[1
]}

Kwak, Joon Young ^{[1
]}

Park, Jongkil ^{[1
]}

Lee, Suyoun ^{[1
]}

Kim, Inho ^{[1
]}

Park, Jong-Keuk ^{[1
]}

Jeong, YeonJoo ^{[1
]}

机构：

[1] Korea Inst Sci & Technol, Ctr Neuromorph Engn, Seoul, South Korea

来源：

FRONTIERS IN COMPUTATIONAL NEUROSCIENCE | 2021年 / 15卷

基金：

新加坡国家研究基金会;

关键词：

spiking neural network; memristor; non-linearity; homeostasis; LTP; LTD ratio; CLASSIFICATION; OPTIMIZATION; NEURONS; DEVICES;

D O I：

10.3389/fncom.2021.646125

中图分类号：

Q [生物科学];

学科分类号：

07 ; 0710 ; 09 ;

摘要：

Among many artificial neural networks, the research on Spike Neural Network (SNN), which mimics the energy-efficient signal system in the brain, is drawing much attention. Memristor is a promising candidate as a synaptic component for hardware implementation of SNN, but several non-ideal device properties are making it challengeable. In this work, we conducted an SNN simulation by adding a device model with a non-linear weight update to test the impact on SNN performance. We found that SNN has a strong tolerance for the device non-linearity and the network can keep the accuracy high if a device meets one of the two conditions: 1. symmetric LTP and LTD curves and 2. positive non-linearity factors for both LTP and LTD. The reason was analyzed in terms of the balance between network parameters as well as the variability of weight. The results are considered to be a piece of useful prior information for the future implementation of emerging device-based neuromorphic hardware.

引用

页数：9

共 39 条

[1]

Agarwal S, 2016, IEEE IJCNN, P929, DOI 10.1109/IJCNN.2016.7727298

[2] Deep Learning with Convolutional Neural Networks Applied to Electromyography Data: A Resource for the Classification of Movements for Prosthetic Hands [J].

Atzori, Manfredo ;

Cognolato, Matteo ;

Mueller, Henning .

FRONTIERS IN NEUROROBOTICS, 2016, 10

[3] Extended memory lifetime in spiking neural networks employing memristive synapses with nonlinear conductance dynamics [J].

Brivio, S. ;

Conti, D. ;

Nair, M., V ;

Frascaroli, J. ;

Covi, E. ;

Ricciardi, C. ;

Indiveri, G. ;

Spiga, S. .

NANOTECHNOLOGY, 2019, 30 (01)

[4] Firing frequency of leaky integrate-and-fire neurons with synaptic current dynamics [J].

Brunel, N ;

Sergi, S .

JOURNAL OF THEORETICAL BIOLOGY, 1998, 195 (01) :87-95

[5] Improving linearity by introducing Al in HfO2 as a memristor synapse device [J].

Chandrasekaran, Sridhar ;

Simanjuntak, Firman Mangasa ;

Saminathan, R. ;

Panda, Debashis ;

Tseng, Tseung-Yuen .

NANOTECHNOLOGY, 2019, 30 (44)

[6] Building Neuromorphic Circuits with Memristive Devices [J].

Chang, Ting ;

Yang, Yuchao ;

Lu, Wei .

IEEE CIRCUITS AND SYSTEMS MAGAZINE, 2013, 13 (02) :56-73

[7] Exploiting the switching dynamics of HfO2-based ReRAM devices for reliable analog memristive behavior [J].

Cueppers, F. ;

Menzel, S. ;

Bengel, C. ;

Hardtdegen, A. ;

von Witzleben, M. ;

Boettger, U. ;

Waser, R. ;

Hoffmann-Eifert, S. .

APL MATERIALS, 2019, 7 (09)

[8] A Survey of Machine Learning Applications for Energy-Efficient Resource Management in Cloud Computing Environments [J].

Demirci, Mehmet .

2015 IEEE 14TH INTERNATIONAL CONFERENCE ON MACHINE LEARNING AND APPLICATIONS (ICMLA), 2015, :1185-1190

[9] Unsupervised learning of digit recognition using spike-timing-dependent plasticity [J].

Diehl, Peter U. ;

Cook, Matthew .

FRONTIERS IN COMPUTATIONAL NEUROSCIENCE, 2015, 9

[10] Neuromorphic Accelerators: A Comparison Between Neuroscience and Machine-Learning Approaches [J].

Du, Zidong ;

Rubin, Daniel D. Ben-Dayan ;

Chen, Yunji ;

He, Liqiang ;

Chen, Tianshi ;

Zhang, Lei ;

Wu, Chengyong ;

Temam, Olivier .

PROCEEDINGS OF THE 48TH ANNUAL IEEE/ACM INTERNATIONAL SYMPOSIUM ON MICROARCHITECTURE (MICRO-48), 2015, :494-507

← 1 2 3 4 →