A mixed-kernel, variable-dimension memristive CNN for electronic nose recognition

Abstract Due to the dynamic characteristics, memristors have great potential for implementing various neural network training and applications. By applying memristors to neural networks as a strategy, the hardware architecture can be taken full advantage of to accelerate matrix operations. A mixed-kernel, variable-dimension memristive convolutional neural network (MixVMCNN), which can identify and classify six gases detected by the electronic nose, is proposed. To generate better gas recognition technique, we provide various receptive fields for the network by using different sizes of convolutional kernels in the same layer. We propose a novel variable-dimension convolutional approach that extracts features in different dimensions. Furthermore, we make use of memristor arrays to make the most of the hardware structure to speed up calculation. Experimental results indicate that the designed network, using only 56 weights, achieves the electronic nose gas classification with high accuracy of 99.72%, based on the Chemical gas sensor array dataset. To further exam the effectiveness of the model, supplemental experiments show that this model has stronger noise-robustness than other networks.

[1]  Jordi Fonollosa,et al.  Chemical gas sensor array dataset , 2015, Data in brief.

[2]  Jiang Wang,et al.  Efficient Spike-Driven Learning With Dendritic Event-Based Processing , 2021, Frontiers in Neuroscience.

[3]  K. Loparo,et al.  BiCoSS: Toward Large-Scale Cognition Brain With Multigranular Neuromorphic Architecture , 2021, IEEE Transactions on Neural Networks and Learning Systems.

[4]  Zhen Sun,et al.  Fault Detection Using the Clustering-kNN Rule for Gas Sensor Arrays , 2016, Sensors.

[5]  T. Masquelier,et al.  S4NN: temporal backpropagation for spiking neural networks with one spike per neuron , 2019, Int. J. Neural Syst..

[6]  Hananel Hazan,et al.  Locally Connected Spiking Neural Networks for Unsupervised Feature Learning , 2019, Neural Networks.

[7]  Bin Deng,et al.  Real-Time Neuromorphic System for Large-Scale Conductance-Based Spiking Neural Networks , 2019, IEEE Transactions on Cybernetics.

[8]  Abbes Amira,et al.  MLP Neural Network Based Gas Classification System on Zynq SoC , 2016, IEEE Access.

[9]  Farnood Merrikh-Bayat,et al.  Training and operation of an integrated neuromorphic network based on metal-oxide memristors , 2014, Nature.

[10]  L. Chua Memristor-The missing circuit element , 1971 .

[11]  Jun Wu,et al.  3D Convolutional Neural Network based on memristor for video recognition , 2020, Pattern Recognit. Lett..

[12]  H.-S. Philip Wong,et al.  Face classification using electronic synapses , 2017, Nature Communications.

[13]  Z. Galias,et al.  Modeling Sinusoidally Driven Self-Directed Channel Memristors , 2018, 2018 International Conference on Signals and Electronic Systems (ICSES).

[14]  Pai Peng,et al.  Gas Classification Using Deep Convolutional Neural Networks , 2018, Sensors.

[15]  Lidan Wang,et al.  Implementation of circuit for reconfigurable memristive chaotic neural network and its application in associative memory , 2020, Neurocomputing.

[16]  Mark Barnell,et al.  Three-dimensional memristor circuits as complex neural networks , 2020, Nature Electronics.

[17]  Ru Huang,et al.  Brain-inspired computing with memristors: Challenges in devices, circuits, and systems , 2020 .

[18]  J.F. Kang,et al.  RRAM based convolutional neural networks for high accuracy pattern recognition and online learning tasks , 2017, 2017 Silicon Nanoelectronics Workshop (SNW).

[19]  Uri C. Weiser,et al.  TEAM: ThrEshold Adaptive Memristor Model , 2013, IEEE Transactions on Circuits and Systems I: Regular Papers.

[20]  A. Bermak,et al.  Fast and robust gas identification system using an integrated gas sensor technology and Gaussian mixture models , 2005, IEEE Sensors Journal.

[21]  Chih-Cheng Chen,et al.  An Improved Fault Diagnosis Using 1D-Convolutional Neural Network Model , 2020, Electronics.

[22]  L.O. Chua,et al.  Memristive devices and systems , 1976, Proceedings of the IEEE.

[23]  Akshay Ananthakrishnan,et al.  All-Passive Hardware Implementation of Multilayer Perceptron Classifiers , 2020, IEEE Transactions on Neural Networks and Learning Systems.

[24]  Jiang Wang,et al.  CerebelluMorphic: Large-Scale Neuromorphic Model and Architecture for Supervised Motor Learning , 2021, IEEE Transactions on Neural Networks and Learning Systems.

[25]  Kristy A. Campbell Self-directed channel memristor for high temperature operation , 2017, Microelectron. J..

[26]  Hongwei Yao,et al.  Corrigendum: Adenylate kinase hCINAP determines self-renewal of colorectal cancer stem cells by facilitating LDHA phosphorylation , 2017, Nature Communications.

[27]  D. Stewart,et al.  The missing memristor found , 2008, Nature.

[28]  John J. Shynk,et al.  Performance surfaces of a single-layer perceptron , 1990, IEEE Trans. Neural Networks.

[29]  Bin Deng,et al.  Scalable Digital Neuromorphic Architecture for Large-Scale Biophysically Meaningful Neural Network With Multi-Compartment Neurons , 2020, IEEE Transactions on Neural Networks and Learning Systems.

[30]  Bin Gao,et al.  Fully hardware-implemented memristor convolutional neural network , 2020, Nature.

[31]  Victor B. Kazantsev,et al.  Spatial Properties of STDP in a Self-Learning Spiking Neural Network Enable Controlling a Mobile Robot , 2020, Frontiers in Neuroscience.

[32]  Leon O. Chua Resistance switching memories are memristors , 2011 .