基于孪生网络的电磁目标跨模式识别算法

doi:10.12263/DZXB.20210805

电子学报 ›› 2022, Vol. 50 ›› Issue (6): 1281-1290.DOI: 10.12263/DZXB.20210805

所属专题：长摘要论文；电磁频谱智能+

• 电磁频谱智能+ • 下一篇

基于孪生网络的电磁目标跨模式识别算法

张伟¹^,², 王沙飞³, 林静然¹^,⁴(), 利强¹^,⁴, 邵怀宗¹^,⁴

^1.电子科技大学信息与通信工程学院，四川成都 611731
^2.电子信息控制重点实验室，四川成都 610036
^3.北方电子设备研究所，北京 100191
^4.鹏城实验室，广东深圳 518055

收稿日期:2021-06-29 修回日期:2021-12-23 出版日期:2022-06-25
- 通讯作者:
- 林静然
- 作者简介:
- 张伟男，1985年出生，山东泰安人.高级工程师，博士研究生.主要研究方向为非合作通信信号处理、电子对抗.
  王沙飞男，1964年出生，河北张家口人.研究员，博士生导师.主要研究方向为非合作通信信号处理、电子对抗.
  林静然（通讯作者）男，1978年出生，重庆人.教授，博士生导师.主要研究方向为电磁频谱智能管控与优化.
  利强男，1982年出生，四川成都人.副教授，博士生导师.主要研究方向为无线通信信号处理与优化.
  邵怀宗男，1969年出生，四川巴中人.教授，博士生导师.主要研究方向为通信和雷达信号处理.
- 基金资助:
- 国家自然科学基金 (U19B2028)

Cross-Modal Recognition Algorithm of Electromagnetic Targets via Siamese Network

ZHANG Wei¹^,², WANG Sha-fei³, LIN Jing-ran¹^,⁴(), LI Qiang¹^,⁴, SHAO Huai-zong¹^,⁴

^1.School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, Sichuan 611731, China
^2.Science and Technology on Electronic Information Control Laboratory, Chengdu, Sichuan 610036, China
^3.Northern Institute of Electronic Equipment of China, Beijing 100191, China
^4.Peng Cheng Laboratory, Shenzhen, Guangdong 518055, China

Received:2021-06-29 Revised:2021-12-23 Online:2022-06-25 Published:2022-06-25
- Corresponding author:
- LIN Jing-ran
- Supported by:
- National Natural Science Foundation of China (U19B2028)

摘要/Abstract

摘要：

以深度学习为代表的人工智能技术是解决电磁目标识别问题的一种有效方法.然而，在识别多模式电磁目标时，目标内部不同模式间数据的差异可能掩盖目标个体间的差异，当某种模式训练样本缺失或稀少时，该模式下的目标识别性能会显著下降.为此，提出一种基于孪生网络的电磁目标跨模式识别算法，在度量学习框架下通过优化设计网络结构和损失函数，引导网络在分类学习过程中拉近同一目标各模式数据间的距离，拉远不同目标数据间的距离，并结合邻近判决准则实现多模式电磁目标在非均衡数据集上的跨模式识别.基于实际数据的测试结果表明，在相同数据集和网络规模条件下，所提方法的跨模式识别率较经典卷积神经网络方法和数据增强方法提升20%.

长摘要
以深度学习为代表的人工智能技术是解决电磁目标识别问题的一种有效方法。然而，现有电磁目标设计日益复杂，大多可以灵活工作在多种差异巨大的模式上，这给目标识别带来新挑战。在识别多模式电磁目标时，目标内部不同模式间数据的差异可能掩盖目标个体间的差异，当某种模式训练样本缺失或稀少时，该模式下的目标识别性能会显著下降，甚至会将工作在不同模式的同一目标误判为多个目标。为此，提出一种基于孪生网络的电磁目标跨模式识别算法，在度量学习框架下通过优化设计网络结构和损失函数，引导网络在分类学习过程中拉近同一目标各模式数据间的距离，拉远不同目标数据间的距离，已达到提取不同模式间共有特征的目的，并结合邻近判决准则实现多模式电磁目标在非均衡数据集上的跨模式识别。在此基础上，进一步设计了联合判决反馈机制，当罕见模式下的测试数据到来时，利用多样本进行联合判决，根据当前判决结果标记新到数据，并利用新标注的数据反馈和更新网络，使网络具备自增长能力，逐渐提升罕见模式下的识别性能。基于实际数据的测试结果表明，在相同数据集和网络规模条件下，所提方法的跨模式识别率较经典卷积神经网络方法和数据增强方法提升20%。

关键词: 电磁目标, 跨模式识别, 孪生网络, 度量学习, 非均衡数据集

Abstract:

The artificial intelligence technology(e.g., deep learning) is an effective approach to electromagnetic target(ET) recognition. However, in the recognition of multi-mode ETs, when the training samples with a certain mode are missing or rare, the recognition rate with this mode will be significantly degraded. The reason mainly lies in that the data distance between different modes of the same ET may exceed the data distance between different ETs. To remedy this, a cross-modal ET recognition approach via Siamese network is developed in this paper. Following the framework of metric learning, we design the network structure and the loss function carefully, so that the recognition training process intentionally drives the Siamese network to enlarge the data distance between different ETs while shorten the distance between different modes of the same ET. Consequently, the multi-mode ETs can still be successfully recognized by employing certain distance-based decision criterion, even with imbalanced training data sets for different modes. The numerical results based on realistic data show that with same data sets and network size, the cross-modal recognition rate of the proposed approach is 20% higher than that of the classical convolutional neural network approach, and that of the popular data-enhancement approaches.

Extended Abstract
The artificial intelligence technology (e.g., deep learning) is an effective approach to electromagnetic target (ET) recognition. However, the design of current ETs is increasingly complex, and most of them can flexibly work on a variety of modes with huge differences, which brings new challenges to ET recognition. Specifically, in the recognition of multi-mode ETs, when the training samples with a certain mode are missing or rare, the recognition rate with this mode will be significantly degraded, or the same target working in different modes may even be mistakenly recognized as multiple targets. The reason mainly lies in that the data distance between different modes of the same ET may exceed the data distance between different ETs. To remedy this, a cross-modal ET recognition approach via Siamese network is developed in this paper. Following the framework of metric learning, we design the network structure and the loss function carefully, so that the recognition training process intentionally drives the Siamese network to enlarge the data distance between different ETs while shorten the distance between different modes of the same ET. As a result, the common features among the signals with different modes of the same target can be extracted, and hence the multi-mode ETs can still be successfully recognized by employing certain distance-based decision criterion, even with imbalanced training data sets for different modes. On this basis, a joint-decision-based feedback mechanism is further designed. When the test data with the rare mode arrives, multiple samples are used for joint decision. Then, the new-coming data is marked according to the joint decision results, and the network is re-trained with the new-labeled data. Consequently, the network gradually improves the recognition performance with the rare mode. The numerical results based on realistic data show that with same data sets and network size, the cross-modal recognition rate of the proposed approach is 20% higher than that of the classical convolutional neural network approach, and that of the popular data-enhancement approaches.

Key words: electromagnetic targets, cross-modal recognition, Siamese network, metric learning, imbalanced data set

中图分类号:

TN911

张伟, 王沙飞, 林静然, 利强, 邵怀宗. 基于孪生网络的电磁目标跨模式识别算法[J]. 电子学报, 2022, 50(6): 1281-1290.

ZHANG Wei, WANG Sha-fei, LIN Jing-ran, LI Qiang, SHAO Huai-zong. Cross-Modal Recognition Algorithm of Electromagnetic Targets via Siamese Network[J]. Acta Electronica Sinica, 2022, 50(6): 1281-1290.

图/表 10

图1 现有深度学习方法在非均衡数据集上的跨模式识别过程

图2 孪生神经网络结构图

图3 孪生网络的Q邻近点判决算法示意图

图4 基于联合判决反馈的自增长跨模式识别原理

表1 单体网络和孪生网络的配置对比

参数	单体网络	孪生网络
网络模型	Resnet50	Resnet50×2
特征提取算法	STFT	STFT
单样本数据长度	10 000	10 000
单样本持续时间	2.6 ms	2.6 ms
单常见模式训练样本数	1 920	1 920
单罕见模式训练样本数(可配)	0/192/384/576/768/1 344/1 920	0/192/384/576/768/1 344/1 920
单罕见模式测试样本数	300	300

图5 罕见模式训练样本完全缺失时的混淆矩阵

图6 第3号、5号电台增加10%的罕见模式训练样本后的混淆矩阵

图7 罕见模式识别率随罕见模式训练样本数量的变化情况

表2 联合判决样本量与识别率的关系

联合判决样本量	联合判决识别率
2	65%
5	72%
20	74%
50	80%
70	91%
100	99.99%

图8 罕见模式识别率随罕见模式训练样本数量的变化情况

参考文献 43

1	郭桂蓉, 庄钊文, 陈曾平. 电磁特征抽取与目标识别[M]. 长沙: 国防科技大学出版社, 1996.
2	王强, 宋京民, 胡建平, 等. 一种快速模板匹配目标识别算法[J]. 计算机工程与应用, 2000, 36(6): 42-43, 69.
	WANGQ, SONGJ M, HUJ P, et al. A improved template matching algorithm of target recognition[J]. Computer Engineering and Applications, 2000, 36(6): 42-43, 69. (in Chinese)
3	郭俊锋, 李言俊, 张科. 利用小波变换进行空间目标电磁特征提取与识别[J]. 火力与指挥控制, 2008, 33(2): 13-15, 20.
	GUOJ F, LIY J, ZHANGK. Study on space target electromagnetic signature extraction and recognition using wavelet transform[J]. Fire Control and Command Control, 2008, 33(2): 13-15, 20. (in Chinese)
4	张量, 刘洋, 王飞, 等. 基于电磁散射特征参数提取与数据分布分析的非线性目标识别[J]. 电波科学学报, 2019, 34(1): 52-59.
	ZHANGL, LIUY, WANGF, et al. Nonlinear target recognition based on electromagnetic scattering feature parameter extraction and data distribution analysis[J]. Chinese Journal of Radio Science, 2019, 34(1): 52-59. (in Chinese)
5	尹宝才, 王文通, 王立春. 深度学习研究综述[J]. 北京工业大学学报, 2015, 41(1): 48-59.
	YINB C, WANGW T, WANGL C. Review of deep learning[J]. Journal of Beijing University of Technology, 2015, 41(1): 48-59. (in Chinese)
6	ZHENGS L, CHENS C, YANGL F, et al. Big data processing architecture for radio signals empowered by deep learning: Concept, experiment, applications and challenges[J]. IEEE Access, 2018, 6: 55907-55922.
7	鲜佩, 张晓芸, 高昭昭. 基于CNN的电磁辐射源目标识别算法[J]. 电子信息对抗技术, 2020, 35(2): 34-38.
	XIANP, ZHANGX Y, GAOZ Z. Electromagnetic radiation source target recognition algorithm based on CNN[J]. Electronic Information Warfare Technology, 2020, 35(2): 34-38. (in Chinese)
8	GAMAF, MARQUESA G, LEUSG, et al. Convolutional neural network architectures for signals supported on graphs[J]. IEEE Transactions on Signal Processing, 2019, 67(4): 1034-1049.
9	田壮壮, 占荣辉, 胡杰民, 等. 基于卷积神经网络的SAR图像目标识别研究[J]. 雷达学报, 2016, 5(3): 320-325.
	TIANZ Z, ZHANR H, HUJ M, et al. SAR ATR based on convolutional neural network[J]. Journal of Radars, 2016, 5(3): 320-325. (in Chinese)
10	O'SHEAT J, ROYT, CLANCYT C. Over-the-air deep learning based radio signal classification[J]. IEEE Journal of Selected Topics in Signal Processing, 2018, 12(1): 168-179.
11	徐彬, 陈渤, 刘家麒, 等. 采用双向LSTM模型的雷达HRRP目标识别[J]. 西安电子科技大学学报, 2019, 46(2): 29-34.
	XUB, CHENB, LIUJ Q, et al. Radar HRRP target recognition by the bidirectional LSTM model[J]. Journal of Xidian University, 2019, 46(2): 29-34. (in Chinese)
12	李东瑾, 杨瑞娟, 李晓柏, 等. 基于核协同表示与鉴别投影的辐射源调制识别[J]. 电子学报, 2020, 48(9): 1695-1702.
	LID J, YANGR J, LIX B, et al. Emitter signal modulation recognition based on kernel collaborative representation and discriminative projection[J]. Acta Electronica Sinica, 2020, 48(9): 1695-1702. (in Chinese)
13	秦鑫, 黄洁, 查雄, 等. 基于扩张残差网络的雷达辐射源信号识别[J]. 电子学报, 2020, 48(3): 456-462.
	QINX, HUANGJ, ZHAX, et al. Radar emitter signal recognition based on dilated residual network[J]. Acta Electronica Sinica, 2020, 48(3): 456-462. (in Chinese)
14	李东瑾, 杨瑞娟, 李晓柏, 等. 基于栈式稀疏降噪自编码网络的辐射源调制识别[J]. 电子学报, 2020, 48(6): 1198-1204.
	LID J, YANGR J, LIX B, et al. Emitter signal modulation recognition based on stacked sparse denoising auto-encoders[J]. Acta Electronica Sinica, 2020, 48(6): 1198-1204. (in Chinese)
15	蔡晶烨. 多模式数字通信电台中的调制解扩解调模块实现[D]. 成都: 电子科技大学, 2008.
	CAIJ Y. Implementation of Modulation, Dispreading, and Demodulation Modules in Multi-Mode Digital Radio Transmission Broadcasting Station[D]. Chengdu: University of Electronic Science and Technology of China, 2008. (in Chinese)
16	阳榴, 朱卫纲, 吕守业, 等. 多功能雷达工作模式识别方法综述[J]. 电讯技术, 2020, 60(11): 1384-1390.
	YANGL, ZHUW G, LYUS Y, et al. Review of multi-function radar mode identification methods[J]. Telecommunication Engineering, 2020, 60(11): 1384-1390. (in Chinese)
17	王沙飞, 李岩, 徐迈. 认知电子战原理与技术[M]. 北京: 国防工业出版社, 2018.
	WANGS F, LIY, XUM. Principle and Technology of Cognitive Electronic Warfare[M]. Beijing: National Defense Industry Press, 2018. (in Chinese)
18	HEH B, GARCIAE A. Learning from imbalanced data[J]. IEEE Transactions on Knowledge and Data Engineering, 2009, 21(9): 1263-1284.
19	KRAWCZYKB. Learning from imbalanced data: Open challenges and future directions[J]. Progress in Artificial Intelligence, 2016, 5(4): 221-232.
20	钱云. 非均衡数据分类算法若干应用研究[D]. 长春: 吉林大学, 2014.
	QIANY. Research on Application of Classification Algorithms for Imbalanced Data[D]. Changchun: Jilin University, 2014. (in Chinese)
21	CHAWLAN V, BOWYERK W, HALLL O, et al. SMOTE: synthetic minority over-sampling technique[J]. Journal of Artificial Intelligence Research, 2002, 16: 321-357.
22	STEFANOWSKIJ. Dealing with Data Difficulty Factors While Learning from Imbalanced Data Challenges in Computational Statistics and Data Mining[M]. Switzerland: Springer, 2016: 333-363.
23	冯宏伟, 姚博, 高原, 等. 基于边界混合采样的非均衡数据处理算法[J]. 控制与决策, 2017, 32(10): 1831-1836.
	FENGH W, YAOB, GAOY, et al. Imbalanced data processing algorithm based on boundary mixed sampling[J]. Control and Decision, 2017, 32(10): 1831-1836. (in Chinese)
24	包萍, 刘运节. 不均衡数据集下基于生成对抗网络的改进深度模型故障识别研究[J]. 电子测量与仪器学报, 2019, 33(3): 176-183.
	BAOP, LIUY J. Research on fault identification based on improved deep model in combination of generative adversarial networks under unbalanced data sets[J]. Journal of Electronic Measurement and Instrumentation, 2019, 33(3): 176-183. (in Chinese)
25	谷琼, 袁磊, 宁彬, 等. 一种基于混合重取样策略的非均衡数据集分类算法[J]. 计算机工程与科学, 2012, 34(10): 128-134.
	GUQ, YUANL, NINGB, et al. A novel classification algorithm for imbalanced datasets based on hybrid resampling strategy[J]. Computer Engineering & Science, 2012, 34(10): 128-134. (in Chinese)
26	ZHOUZ H, LIUX Y. On multi-class cost-sensitive learning[J]. Computational Intelligence, 2010, 26(3): 232-257.
27	谷琼, 袁磊, 熊启军, 等. 基于非均衡数据集的代价敏感学习算法比较研究[J]. 微电子学与计算机, 2011, 28(8): 146-149, 153.
	GUQ, YUANL, XIONGQ J, et al. A comparative study of cost-sensitive learning algorithm based on imbalanced data sets[J]. Microelectronics & Computer, 2011, 28(8): 146-149, 153. (in Chinese)
28	JAPKOWICZN, MYERSC, GLUCKM. A novelty detection approach to classification//Proceedings of the 14th International Joint Conference on Artificial Intelligence. Montreal: Morgan Kaufmann Publishers Inc, 1995: 518-523.
29	KRAWCZYKB, WOŹNIAKM, HERRERAF. On the usefulness of one-class classifier ensembles for decomposition of multi-class problems[J]. Pattern Recognition, 2015, 48(12): 3969-3982.
30	张旭, 周新志, 赵成萍, 等. 基于犹豫模糊决策树的非均衡数据分类[J]. 计算机工程, 2019, 45(8): 75-79, 91.
	ZHANGX, ZHOUX Z, ZHAOC P, et al. Unbalanced data classification based on hesitant fuzzy decision tree[J]. Computer Engineering, 2019, 45(8): 75-79, 91. (in Chinese)
31	钟瑛, 朱顺痣, 曾志强, 等. 一种基于核学习的非均衡数据分类算法[J]. 厦门大学学报(自然科学版), 2012, 51(2): 189-194.
	ZHONGY, ZHUS Z, ZENGZ Q, et al. A classfication method for imbalance dataset based on kernel learning[J]. Journal of Xiamen University (Natural Science), 2012, 51(2): 189-194. (in Chinese)
32	WANGZ, YEX J, WANGC K, et al. Network embedding with completely-imbalanced labels[J]. IEEE Transactions on Knowledge and Data Engineering, 2021, 33(11): 3634-3647.
33	李卓然, 张永. 基于集成的非均衡数据分类主动学习算法[J]. 计算机应用与软件, 2012, 29(6): 81-83, 88.
	LIZ R, ZHANGY. Imbalanced data classification active learning algorithm based on boosting[J]. Computer Applications and Software, 2012, 29(6): 81-83, 88. (in Chinese)
34	WOŹNIAKM, GRAÑAM, CORCHADOE. A survey of multiple classifier systems as hybrid systems[J]. Information Fusion, 2014, 16: 3-17.
35	WANGS Z, LIZ J, CHAOW H, et al. Applying adaptive over-sampling technique based on data density and cost-sensitive SVM to imbalanced learning[C]//The 2012 International Joint Conference on Neural Networks(IJCNN). Brisbane: IEEE, 2012: 1-8.
36	LIUS J, SHIQ, ZHANGL P. Few-shot hyperspectral image classification with unknown classes using multitask deep learning[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 59(6): 5085-5102.
37	DONGY H, JIANGX H, ZHOUH J, et al. SR2CNN: zero-shot learning for signal recognition[J]. IEEE Transactions on Signal Processing, 2021, 69: 2316-2329.
38	徐丹蕾, 杜兰, 王鹏辉, 等. 采用多任务稀疏学习的雷达HRRP小样本目标识别[J]. 西安电子科技大学学报, 2016, 43(2): 23-28.
	XUD L, DUL, WANGP H, et al. Radar HRRP target recognition by utilizing multitask sparse learning with a small training data size[J]. Journal of Xidian University, 2016, 43(2): 23-28. (in Chinese)
39	兰红, 方治屿. 零样本图像识别[J]. 电子与信息学报, 2020, 42(5): 1188-1200.
	LANH, FANGZ Y. Recent advances in zero-shot learning[J]. Journal of Electronics & Information Technology, 2020, 42(5): 1188-1200. (in Chinese)
40	ZHOUH J, JIAOL C, ZHENGS L, et al. Generative adversarial network-based electromagnetic signal classification: A semi-supervised learning framework[J]. China Communications, 2020, 17(10): 157-169.
41	DONGY H, JIANGX H, CHENGL, et al. SSRCNN: A semi-supervised learning framework for signal recognition[J]. IEEE Transactions on Cognitive Communications and Networking, 2021, 7(3): 780-789.
42	TAOR, GAVVESE, SMEULDERSA W M. Siamese instance search for tracking[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 1420-1429.
43	WANGX, HANX T, HUANGW L, et al. Multi-similarity loss with general pair weighting for deep metric learning[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Long Beach: IEEE, 2019: 5017-5025.

[1]	付利华, 赵宇, 姜涵煦, 赵茹, 吴会贤, 闫绍兴. 基于前景感知视觉注意的半监督视频目标分割[J]. 电子学报, 2022, 50(1): 195-206.
[2]	李琳, 唐守廉. 融合内容表示的度量排序学习推荐模型[J]. 电子学报, 2020, 48(8): 1615-1622.
[3]	付利华, 赵宇, 孙晓威, 卢中山, 王丹, 杨寒雪. 基于孪生网络的快速视频目标分割[J]. 电子学报, 2020, 48(4): 625-630.

基于孪生网络的电磁目标跨模式识别算法

Cross-Modal Recognition Algorithm of Electromagnetic Targets via Siamese Network

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 43

相关文章 3

编辑推荐

Metrics