面向声呐图像水下目标分类的类脑SNN研究

doi:10.3778/j.issn.1002-8331.2203-0003

摘要/Abstract

摘要： 声呐图像被广泛应用于复杂海况的水下救援和海底探测中，长时的人工搜索极易造成视觉疲劳而错失目标。无人潜航器可大幅降低搜索工作量和主观误差，但这取决于无人自主系统的能效和自动分类性能。卷积神经网络的训练和推理需要比较高的能耗，难以在无人潜航器的移动环境下部署和应用，而且声呐图像训练数据稀少和样本不平衡也增加了模型训练的难度。脉冲神经网络通过二进制离散的时序脉冲信号可以避免卷积神经网络中高昂的乘法计算代价，具有低能耗和高精度的特性。构建了可用于合成孔径声呐图像分类的浅层脉冲神经网络，设计了一种基于脉冲神经网络的小样本水下目标分类算法。采用基于风格迁移的模拟声呐图像生成方法和加权随机采样方法，缓解了声呐图像训练数据稀少和样本不平衡问题。实验表明，在声呐图像样本稀少和不平衡的情况下，算法的分类准确率高于ResNet50、VGG19和MobileNet V2等架构的卷积神经网络，达到91.11%。计算复杂度和能耗分析也表明，脉冲神经网络相比于卷积神经网络具有很大优势。脉冲神经网络是研究和实现类脑计算非常合适的模型，可满足无人水下航行器的移动计算需求，该研究对实现无人自主设备的智能应用具有先进的技术优势。

关键词: 脉冲神经网络, 近似求导的反向传播算法, 合成孔径声呐图像, 侧扫声呐, 水下目标分类

Abstract: Sonar images are widely used in underwater rescue and seabed exploration under complex sea conditions. Long-term manual search is very easy to cause visual fatigue and lost the target. Unmanned underwater vehicle can greatly reduce the workload and subjective error, but it depends on energy efficiency and automatic classification performance of unmanned autonomous system. The training and inference of convolutional neural network need high energy consumption, thus the conventional deep neural network is difficult to deploy and apply in the mobile environment of unmanned underwater vehicle, the scarcity of sonar image training data and imbalanced samples also increase the difficulty of model training. Spiking neural network can avoid the high multiplication cost in convolutional neural network through binary discrete timing spiking signal, and has the characteristics of low energy consumption and high-precision. In this paper, a low-energy shallow spiking neural network for synthetic aperture sonar image classification is constructed, and a small sample target classification algorithm based on spiking neural network is designed. The simulated sonar image generation method based on style transfer and weighted random sampling method are adopted to alleviate the problems of scarce sonar image training data and sample imbalance. Experiments show that when the sonar image samples are scarce and unbalanced, the classification accuracy of the algorithm is higher than that of convolutional neural networks based on ResNet50, VGG19 and MobileNet V2, up to 91.11%. The analysis of computational complexity and energy consumption shows that spiking neural network has great advantages over convolutional neural network. Spiking neural network is a very suitable model for the research and implementation of brain-inspired computing, and it can meet the requirements of mobile computing of unmanned underwater vehicles. This research has advanced technical advantages for the intelligent application of unmanned autonomous equipment.

Key words: spiking neural network, approximate derivation backpropagation, synthetic aperture sonar image, sidescan sonar, underwater target classification

刘扬, 田猛, 曹珂境, 王瑞毅, 赵伟. 面向声呐图像水下目标分类的类脑SNN研究[J]. 计算机工程与应用, 2023, 59(10): 204-212.

LIU Yang, TIAN Meng, CAO Kejing, WANG Ruiyi, ZHAO Wei. Research on Brain-Inspired SNN for Underwater Target Classification of Sonar Images[J]. Computer Engineering and Applications, 2023, 59(10): 204-212.

参考文献

[1] KUMAR N，TAN Q F，NARAYANAN S S.Object classification in sidescan sonar images with sparse representation techniques[C]//Proceedings of the 2012 IEEE International Conference on Acoustics，Speech and Signal Processing，Kyoto，Mar 25-30，2012.Piscataway：IEEE，2012：1333-1336.
[2] RHINELANDER J.Feature extraction and target classification of side-scan sonar images[C]//Proceedings of the 2016 IEEE Symposium Series on Computational Intelligence，Athens，Dec 6-9，2016.Piscataway：IEEE，2017：1-6.
[3] ZHU M，SONG Y，GUO J，et al.PCA and kernel-based extreme learning machine for side-scan sonar image classification[C]//Proceedings of the 2017 IEEE Underwater Technology，Busan，Feb 21-24，2017.Piscataway：IEEE，2017：1-5.
[4] HUO G，WU Z，LI J.Underwater object classification in sidescan sonar images using deep transfer learning and semisynthetic training data[J].IEEE Access，2020，8：47407-47418.
[5] GE Q，RUAN F，QIAO B，et al.Side-scan sonar image classification based on style transfer and pre-trained convolutional neural networks[J].Electronics，2021，10（15）：1823.
[6] MAASS W.Networks of spiking neurons：The third generation of neural network models[J].Neural Networks，1997，10（9）：1659-1671.
[7] DENG L，WU Y，HU X，et al.Rethinking the performance comparison between SNNS and ANNS[J].Neural Networks，2019，121：294-307.
[8] PAINKRAS E，PLANA L A，GARSIDE J，et al.SpiNNaker：a 1-W 18-core system-on-chip for massively-parallel neural network simulation[J].IEEE Journal of Solid-State Circuits，2013，48（8）：1943-1953.
[9] FURBER S B，GALLUPPI F，TEMPLE S，et al.The SpiNNaker project[J].Proceedings of the IEEE，2014，102（5）：652-665.
[10] BENJAMIN B V，GAO P，MCQUINN E，et al.Neurogrid：a mixed-analog-digital multichip system for large-scale neural simulations[J].Proceedings of the IEEE，2014，102（5）：699-716.
[11] 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.
[12] DAVIES M，SRINIVASA N，LIN T H，et al.Loihi：a neuromorphic manycore processor with on-chip learning[J].IEEE Micro，2018，38（1）：82-99.
[13] 郁龚健，张鲁飞，李佩琦，等.SWAM：SNN工作负载自动映射器[J].计算机科学与探索，2021，15（9）：1641-1657.
YU G J，ZHANG L F，LI P Q，et al.SWAM：workload automatic mapper for SNN[J].Journal of Frontiers of Computer Science and Technology，2021，15（9）：1641-1657.
[14] 李佩琦，郁龚健，华夏，等.PEST：由PYNQ集群实现的高能效NEST类脑仿真器[J].计算机科学与探索，2021，15（11）：2127-2141.
LI P Q，YU G J，HUA X，et al.PEST：energy-efficient NEST brain-like simulator implemented by PYNQ cluster[J].Journal of Frontiers of Computer Science and Technology，2021，15（11）：2127-2141.
[15] DAYAN P，ABBOTT L.Theoretical neuroscience[M].Cambridge：MIT Press，2001：162-166.
[16] IZHIKEVICH，EUGENE M.Simple model of spiking neurons[J].IEEE Transactions on Neural Networks，2003，14（16）：1569-1572.
[17] HODGKIN A L，HUXLEY A F.A quantitative description of membrane current and its application to conduction and excitation in nerve[J].The Journal of Physiology，1952，117（4）：500-544.
[18] KORNIJCUK V，LIM H，SEOK J Y，et al.Leaky integrate-and-fire neuron circuit based on floating-gate integrator[J].Frontiers in Neuroscience，2016，10：212.
[19] GATYS L A，ECKER A S，BETHGE M.Image style transfer using convolutional neural networks[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition，Las Vegas，Jun 27-30，2016.Piscataway：IEEE，2016：2414-2423.
[20] LEE C，SARWAR S S，PANDA P，et al.Enabling spike-based backpropagation for training deep neural network architectures[J].Frontiers in Neuroscience，2020，14：119.
[21] HAN S，POOL J，TRAN J，et al.Learning both weights and connections for efficient neural network[C]//Advances in Neural Information Processing Systems，Montréal，Dec 7-12，2015：1135-1143.
[22] LIN D D，TALATHI S S，SREEKANTH ANNAPUREDDY V.Fixed point quantization of deep convolutional networks[C]//2016 International Conference on Machine Learning，New York，Jun 19-24，2016.Piscataway：IEEE，2016：2849-2858.
[23] LECUN Y，BOTTOU L，BENGIO Y，et al.Gradient-based learning applied to document recognition[J].Proceedings of the IEEE，1998，86（11）：2278-2324.
[24] SIMONYAN K，ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[J].arXiv：14091556，2014.
[25] HE K，ZHANG X，REN S，et al.Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition，Las Vegas，Jun 27-30，2016.Piscataway：IEEE，2016：770-778.
[26] SANDLER M，HOWARD A，ZHU M，et al.MobileNetV2：inverted residuals and linear bottlenecks[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition，Salt Lake City，Jun 18-23，2018.Piscataway：IEEE，2018：4510-4520.