基于散射特征增强的轻量化虚假目标欺骗干扰判别网络

蒲巍; 伍宇恒; 宋月; 王重淞; 吴万敏; 刘欣远; 武俊杰; 黄钰林; 杨建宇

doi:10.12000/JR25275

基于散射特征增强的轻量化虚假目标欺骗干扰判别网络

DOI: 10.12000/JR25275 CSTR: 32380.14.JR25275

电子科技大学信息与通信工程学院成都 611731

基金项目: 国家自然科学基金(U25B200901)

详细信息

作者简介:
蒲　巍，博士，教授，主要研究方向为智能化雷达成像、 SAR运动补偿等

伍宇恒，硕士生，主要研究方向为雷达智能抗干扰

宋　月，博士，主要研究方向为雷达成像、散射特性分析、提取与利用等

王重淞，博士，主要研究方向为SAR抗干扰、智能信号处理等

吴万敏，博士，主要研究方向为SAR运动目标成像、雷达信号处理等

刘欣远，博士，主要研究方向为抗干扰成像、智能信号处理等

武俊杰，博士，教授，主要研究方向为双/多基SAR、智能化雷达成像等

黄钰林，博士，教授，主要研究方向为雷达探测与成像，雷达目标检测与识别等

杨建宇，博士，教授，主要研究方向为前视雷达成像、双/多基SAR成像、新体制雷达探测与成像等

通讯作者:
蒲巍 puwei@uestc.edu.cn

责任主编：黄岩 Corresponding Editor: HUANG Yan

计量
- 文章访问数:
- HTML全文浏览量:
- PDF下载量:
- 被引次数: 0
出版历程
- 收稿日期: 2025-12-23
- 修回日期: 2026-05-24

A Scattering Feature-Enhanced Network for SAR Deceptive Jamming Discrimination

School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China

Funds: The National Natural Science Foundation of China (U25B200901)

More Information

Corresponding author: Pu Wei, puwei@uestc.edu.cn

摘要

摘要: 合成孔径雷达(Synthetic Aperture Radar, SAR)具备全天时全天候高分辨成像的能力，在监视侦察、防空反导、灾害监测等军民领域发挥着重要的作用。然而，伴随着电子对抗技术的发展，雷达干扰机可对SAR成像结果造成虚假目标欺骗干扰，严重威胁到SAR图像的判读和实时决策。针对上述问题，该文结合目标电磁散射机理，提出了基于散射特征增强的SAR虚假目标欺骗干扰判别网络(SF-ViT)，该网络针对干扰机空间位置固定导致的回波方位分布差异和模板构型与信号参数不同导致的散射特征差异，通过一个浅层特征增强模块放大真实目标和虚假目标在图像域上的区别，再通过卷积-ViTs混合轻量化网络完成高维语义特征的提取和分类。经过本文构建的SAR虚假目标欺骗干扰数据集上的对比试验验证，所提网络在不同信噪比条件下可以达到94.97%的平均判别准确率，同时具有参数量低、易于部署到边缘设备的优势，并且通过消融实验验证了所提散射特征增强模块也可以与传统模型相结合，提高对SAR虚假目标欺骗干扰判别准确率。
- 合成孔径雷达(SAR) /
- 欺骗干扰 /
- 散射特征 /
- 卷积神经网络(CNN) /
- 视觉 Transformer (ViT)
Abstract: Synthetic aperture radar (SAR) enables round-the-clock high-resolution imaging under all weather conditions, thereby playing a vital role in both military domains (e.g., surveillance, reconnaissance, air defense, and missile defense) and civilian domains (e.g., disaster monitoring).. However, advancements in electronic countermeasure technologies have led to the development of radar jammers that generate deceptive jamming with false targets in SAR imagery. This seriously undermines the interpretation of SAR images and real-time decision-making. To tackle these issues, this study proposes a scattering feature–enhanced vision Transformer-based network (SF-ViT) to discriminate deceptive jamming using SAR false targets, which leverages the electromagnetic scattering mechanisms of targets. By targeting the azimuth distribution disparity of echoes caused by the fixed spatial positions of jammers and the scattering feature discrepancy induced by variations in template configurations and signal parameters, the network first highlights the differences between real and false targets in the image domain using a shallow feature enhancement module. Subsequently, it extracts and classifies high-dimensional semantic features through a lightweight hybrid convolutional–ViT network. Experimental validation on the SAR false-target deceptive jamming dataset built in this study indicates that the proposed network attains an average discrimination accuracy of 94.97% under diverse signal-to-noise ratio conditions and requires fewer parameters, making it easy to deploy on edge devices. In addition, ablation experiments demonstrate that the proposed scattering feature enhancement module can be integrated with traditional models, further enhancing the discrimination accuracy of SAR false-target deceptive jamming.
- Synthetic Aperture Radar (SAR) /
- Deceptive Jamming /
- Scattering Feature /
- Convolutional Neural Network /
- XXX

HTML全文

图 1 SAR虚假目标欺骗干扰几何模型

Figure 1. Geometric model of SAR deceptive jamming against false targets

下载: 全尺寸图片幻灯片

图 2 4种干扰算法流程图

Figure 2. Flowcharts of four jamming algorithms

下载: 全尺寸图片幻灯片

图 3 欺骗干扰特性及回波包络、方位分布位置示意图

Figure 3. Schematic diagram of deceptive jamming characteristics, echo envelope, and azimuth distribution

下载: 全尺寸图片幻灯片

图 4 干扰效果示意图

Figure 4. Schematic diagram of jamming effects

下载: 全尺寸图片幻灯片

图 5 成像结果相位示意图

Figure 5. Schematic diagram of the phase in imaging results

下载: 全尺寸图片幻灯片

图 6 SF-ViT判别网络结构示意图

Figure 6. Schematic diagram of the SF-ViT discriminative network structure

下载: 全尺寸图片幻灯片

图 7 卷积加性注意力机制示意图

Figure 7. Schematic diagram of the convolutional additive attention mechanism

下载: 全尺寸图片幻灯片

图 8 部分卷积核示意图

Figure 8. Schematic diagram of partial convolution kernels

下载: 全尺寸图片幻灯片

图 9 散射特征增强模块处理流程示意图

Figure 9. Schematic diagram of the processing flow of the scattering feature enhancement module

下载: 全尺寸图片幻灯片

图 10 基于散射特征增强的SAR干扰目标判别网络的具体结构

Figure 10. Detailed architecture of the SAR jamming target discrimination network based on scattering feature enhancement

下载: 全尺寸图片幻灯片

图 11 MSTAR数据集中的SAR图像及电磁计算仿真图像展示

Figure 11. Illustration of SAR images from the MSTAR dataset and electromagnetic computation simulation images

下载: 全尺寸图片幻灯片

图 12 SAR虚假目标欺骗干扰数据集中的示例图像。第一行为真实目标成像结果，第二行为对应模板的干扰成像结果

Figure 12. Example images from the SAR deceptive jamming dataset. The first row shows imaging results of real targets, while the second row displays corresponding jamming imaging results generated using templates

下载: 全尺寸图片幻灯片

表 1 不同方法对SAR虚假目标欺骗干扰数据集的判别性能

Table 1. Discrimination performance of different methods on the SAR deceptive jamming dataset

信噪比(dB)/模型	resnet50	inception_v4	xception	MCTNAS	GreedyNAS	ProxylessNAS	DARTS	ActGen	ViT	SF-ViT
–15	68.84	72.11	63.82	72.11	77.89	74.62	80.40	84.67	87.94	90.95
–12	73.37	75.88	71.86	75.63	78.39	76.13	82.16	85.93	88.19	92.46
–9	77.89	77.39	75.63	81.41	80.15	79.4	85.18	86.93	88.69	95.23
–6	84.92	83.17	83.92	85.93	84.42	83.42	87.44	88.69	89.45	95.48
–3	90.20	89.95	88.44	90.70	87.94	85.93	89.45	90.95	90.45	96.23
0	90.95	90.95	93.22	94.47	90.95	89.95	91.21	92.46	91.46	96.98
5	94.22	94.47	94.72	95.98	92.96	92.71	92.71	92.96	93.47	97.49
Average	82.91	83.42	81.66	85.18	84.67	83.17	86.94	88.94	89.95	94.97

下载: 导出CSV

表 2 各网络计算复杂度和参数量

Table 2. Computational complexity and parameter count of each network

模型	Flops (G)	Params (M)
resnet50	4.1317	23.51
Inception_v4	6.1536	41.15
xception	4.5974	20.81
MCT-NAS	0.4416	8.45
GreedyNAS	0.3699	6.50
ProxylessNAS	0.3204	4.08
DARTS-V2	0.5299	4.72
ActGen	4.1317	23.51
SF-ViT	0.7977	3.98

下载: 导出CSV

表 3 融合散射特征增强模块的各方法在SAR虚假目标欺骗干扰数据集上的判别性能

Table 3. Discrimination performance of various methods incorporating the scattering feature enhancement module on the SAR deceptive jamming dataset

信噪比(dB)/模型	resnet50	inception_v4	xception	MCTNAS	GreedyNAS	ProxylessNAS	DARTS	ActGen
-15	82.16	80.40	82.91	87.44	84.67	81.66	90.20	89.70
-12	84.92	84.92	86.43	88.94	85.68	83.67	90.45	91.21
-9	87.69	87.94	90.45	89.45	88.19	86.18	91.46	91.71
-6	91.21	90.45	91.71	91.71	89.20	88.44	91.96	93.72
-3	95.23	93.22	95.23	92.96	91.71	90.45	93.47	95.48
0	96.48	95.23	95.98	95.48	94.97	94.47	95.48	96.48
5	97.24	97.49	97.49	97.74	96.98	95.98	95.98	97.74
Average	90.70	89.95	91.46	91.96	90.20	88.69	92.71	93.72

下载: 导出CSV

参考文献(48)

[1]	武俊杰, 杨建宇, 李中余, 等. 双基地SAR成像处理方法综述[J]. 雷达学报(中英文), 2025, 14(5): 1115–1141. doi: 10.12000/JR25067. WU Junjie, YANG Jianyu, LI Zhongyu, et al. Review of bistatic synthetic aperture radar imaging methods[J]. Journal of Radars, 2025, 14(5): 1115–1141. doi: 10.12000/JR25067.
[2]	保铮, 邢孟道, 王彤. 雷达成像技术[M]. 北京: 电子工业出版社, 2005. BAO Zheng, XING Mengdao, and WANG Tong. Radar Imaging Technology[M]. Beijing: Publishing House of Electronics Industry, 2005. (查阅网上资料,未找到本条文献英文翻译信息,请确认).
[3]	CONDLEY C J. Some system considerations for electronic countermeasures to synthetic aperture radar[C]. IEE Colloquium on Electronic Warfare Systems, London, UK, 1991: 8/1–8/7.
[4]	王雪松, 刘建成, 张文明, 等. 间歇采样转发干扰的数学原理[J]. 中国科学 E 辑信息科学, 2006, 36(8): 891–901. doi: 10.1360/zf2006-36-8-891. WANG Xuesong, LIU Jiancheng, ZHANG Wenming, et al. Mathematical principle of intermittent sampling and forward interference[J]. Science in China Series E: Information Sciences, 2006, 36(8): 891–901. doi: 10.1360/zf2006-36-8-891.
[5]	吴晓芳, 王雪松, 卢焕章. 对SAR的间歇采样转发干扰研究[J]. 宇航学报, 2009, 30(5): 2043–2048,2072. doi: 10.3873/j.issn.1000-1328.2009.05.050. WU Xiaofang, WANG Xuesong, and LU Huanzhang. Study of intermittent sampling repeater jamming to SAR[J]. Journal of Astronautics, 2009, 30(5): 2043–2048,2072. doi: 10.3873/j.issn.1000-1328.2009.05.050.
[6]	胡东辉, 吴一戎. 合成孔径雷达散射波干扰研究[J]. 电子学报, 2002, 30(12): 1882–1884. doi: 10.3321/j.issn:0372-2112.2002.12.040. HU Donghui and WU Yirong. The scatter-wave jamming to SAR[J]. Acta Electronica Sinica, 2002, 30(12): 1882–1884. doi: 10.3321/j.issn:0372-2112.2002.12.040.
[7]	王盛利, 于立, 倪晋鳞, 等. 合成孔径雷达的有源欺骗干扰方法研究[J]. 电子学报, 2003, 31(12): 1900–1902. doi: 10.3321/j.issn:0372-2112.2003.12.035. WANG Shengli, YU Li, NI Jinlin, et al. A study on the active deception jamming to SAR[J]. Acta Electronica Sinica, 2003, 31(12): 1900–1902. doi: 10.3321/j.issn:0372-2112.2003.12.035.
[8]	WANG Wenjing, WU Junjie, PEI Jifang, et al. Deception-jamming localization and suppression via configuration optimization for multistatic SAR[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5232016. doi: 10.1109/TGRS.2022.3189409.
[9]	LIU Yongcai, WANG Wei, PAN Xiaoyi, et al. Inverse omega-K algorithm for the electromagnetic deception of synthetic aperture radar[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016, 9(7): 3037–3049. doi: 10.1109/JSTARS.2016.2543961.
[10]	SHI Xiaoran, ZHOU Feng, ZHAO Bo, et al. Deception jamming method based on micro-Doppler effect for vehicle target[J]. IET Radar, Sonar & Navigation, 2016, 10(6): 1071–1079. doi: 10.1049/iet-rsn.2015.0371.
[11]	ZHAO Bo, HUANG Lei, ZHOU Feng, et al. Performance improvement of deception jamming against SAR based on minimum condition number[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2017, 10(3): 1039–1055. doi: 10.1109/JSTARS.2016.2614957.
[12]	刘宁, 赵博, 黄磊. 单通道SAR抗欺骗干扰方法[J]. 雷达学报, 2019, 8(1): 73–81. doi: 10.12000/JR18072. LIU Ning, ZHAO Bo, and HUANG Lei. Anti-deceptive jamming methods based on single-channel synthetic aperture radar[J]. Journal of Radars, 2019, 8(1): 73–81. doi: 10.12000/JR18072.
[13]	LOU Mingyue, YANG Jianyu, LI Zhongyu, et al. Joint optimal and adaptive 2-D spatial filtering technique for FDA-MIMO SAR deception jamming separation and suppression[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5238414. doi: 10.1109/TGRS.2022.3221786.
[14]	ZHOU Bo, LI Yifan, and WANG Zuhang. A fast startup crystal oscillator with digital SAR-AFC based two-step injection[J]. Chinese Journal of Electronics, 2024, 33(5): 1147–1153. doi: 10.23919/cje.2023.00.043.
[15]	BANG Huang, WANG Wenqin, ZHANG Shunsheng, et al. FDA-based space-time-frequency deceptive jamming against SAR imaging[J]. IEEE Transactions on Aerospace and Electronic Systems, 2022, 58(3): 2127–2140. doi: 10.1109/TAES.2021.3130212.
[16]	HUANG Tianyou, LIN Huifu, YANG Chao, et al. A 2-D multiplication modulation jamming method against high-resolution spaceborne SAR based on defocus correction[J]. IEEE Geoscience and Remote Sensing Letters, 2025, 22: 4005805. doi: 10.1109/LGRS.2025.3534222.
[17]	LIU Yangyang, LAN Lan, ZHONG Lei, et al. A deceptive jamming approach for SAR based on range-azimuth modulation[J]. IEEE Geoscience and Remote Sensing Letters, 2025, 22: 4013205. doi: 10.1109/LGRS.2025.3613363.
[18]	赵博, 陈基, 黄磊. 单比特多模态SAR干扰方法研究[J]. 雷达学报, 2022, 11(6): 1119–1130. doi: 10.12000/JR22176. ZHAO Bo, CHEN Ji, and HUANG Lei. One-bit multi-modality jamming method against SAR[J]. Journal of Radars, 2022, 11(6): 1119–1130. doi: 10.12000/JR22176.
[19]	DING Chang, MU Huilin, SHI Yunzhou, et al. Dual-polarized and conformal time-modulated metasurface-based 2-D jamming against SAR imaging systems[J]. IEEE Transactions on Antennas and Propagation, 2025, 73(10): 7752–7764. doi: 10.1109/TAP.2025.3581400.
[20]	WANG Bing, CUI Guolong, ZHANG Shuai, et al. Deceptive jamming suppression based on coherent cancelling in multistatic radar system[C]. 2016 IEEE Radar Conference (RadarConf), Philadelphia, USA, 2016: 1–5. doi: 10.1109/RADAR.2016.7485304.
[21]	ZHAO Bo, HUANG Lei, LI Jian, et al. Target reconstruction from deceptively jammed single-channel SAR[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 56(1): 152–167. doi: 10.1109/TGRS.2017.2744178.
[22]	SHANG Yuanzhe, PU Wei, WU Congwen, et al. HDSS-Net: A novel hierarchically designed network with spherical space classifier for ship recognition in SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 5222420. doi: 10.1109/TGRS.2023.3332137.
[23]	王佳祥, 孟进, 李伟, 等. YOLO-S³: 一种轻量化的雷达复合干扰识别网络[J]. 雷达学报(中英文). doi: 10.12000/JR25080. WANG Jiaxiang, MENG Jin, LI Wei, et al. YOLO-S³: A lightweight network for radar composite jamming signal recognition[J]. Journal of Radars. doi: 10.12000/JR25080.
[24]	张顺生, 陈爽, 陈晓莹, 等. 面向小样本的多模态雷达有源欺骗干扰识别方法[J]. 雷达学报, 2023, 12(4): 882–891. doi: 10.12000/JR23104. ZHANG Shunsheng, CHEN Shuang, CHEN Xiaoying, et al. Active deception jamming recognition method in multimodal radar based on small samples[J]. Journal of Radars, 2023, 12(4): 882–891. doi: 10.12000/JR23104.
[25]	陈思伟, 崔兴超, 李铭典, 等. 基于深度CNN模型的SAR图像有源干扰类型识别方法[J]. 雷达学报, 2022, 11(5): 897–908. doi: 10.12000/JR22143. CHEN Siwei, CUI Xingchao, LI Mingdian, et al. SAR image active jamming type recognition based on deep CNN model[J]. Journal of Radars, 2022, 11(5): 897–908. doi: 10.12000/JR22143.
[26]	SHAO Guangqing, CHEN Yushi, and WEI Yinsheng. Convolutional neural network-based radar jamming signal classification with sufficient and limited samples[J]. IEEE Access, 2020, 8: 80588–80598. doi: 10.1109/ACCESS.2020.2990629.
[27]	陈泽伟, 严远鹏. 基于改进DCGAN的毫米波雷达相互干扰时频图像生成研究——以生成样本对CNN干扰抑制模型性能影响为例[J]. 现代信息科技, 2022, 6(13): 55–61. doi: 10.19850/j.cnki.2096-4706.2022.013.014. CHEN Zewei and YAN Yuanpeng. Research on generation of MMW radar mutual interference time-frequency image based on improved DCGAN-a case of performance effect of the generated samples on the CNN interference suppression model[J]. Modern Information Technology, 2022, 6(13): 55–61. doi: 10.19850/j.cnki.2096-4706.2022.013.014.
[28]	王重淞, 蒲巍, 高杰, 等. 低干信比SAR非虚假目标类有源干扰的轻量化鉴别网络[J]. 雷达学报(中英文). (查阅网上资料,未找到本条文献卷期、页码信息,请确认). doi: 10.12000/JR25195. WANG Chongsong, PU Wei, GAO Jie, et al. Lightweight discrimination network for non-spoofing active jamming in SAR under low JSR[J]. Journal of Radars. doi: 10.12000/JR25195.
[29]	刘永才. 对多模式高分辨合成孔径雷达的大场景欺骗干扰技术研究[D]. [博士论文], 国防科技大学, 2017. doi: 10.27052/d.cnki.gzjgu.2017.000037. LIU Yongcai. Study on large-area deceptive jamming against multi-mode high-resolution synthetic aperture radar[D]. [Ph.D. dissertation], National University of Defense Technology, 2017. doi: 10.27052/d.cnki.gzjgu.2017.000037.
[30]	SONG Yue, PU Wei, ZHANG Yuhua, et al. Analysis of BiSAR images characteristics: From an electromagnetic scattering perspective[J]. IEEE Transactions on Aerospace and Electronic Systems, 2026, 62: 2852–2869. doi: 10.1109/TAES.2025.3641918.
[31]	SONG Yue, PU Wei, ZHANG Yin, et al. A structure-driven multistage trajectory planning method for BiSAR[J]. IEEE Transactions on Geoscience and Remote Sensing, 2025, 63: 2005114. doi: 10.1109/TGRS.2025.3642973.
[32]	邢孟道, 谢意远, 高悦欣, 等. 电磁散射特征提取与成像识别算法综述[J]. 雷达学报, 2022, 11(6): 921–942. doi: 10.12000/JR22232. XING Mengdao, XIE Yiyuan, GAO Yuexin, et al. Electromagnetic scattering characteristic extraction and imaging recognition algorithm: A review[J]. Journal of Radars, 2022, 11(6): 921–942. doi: 10.12000/JR22232.
[33]	赵博. 合成孔径雷达欺骗干扰方法研究[D]. [博士论文], 西安电子科技大学, 2015. ZHAO Bo. Study on methods of synthetic aperture radar deception jamming[D]. [Ph.D. dissertation], Xidian University, 2015.
[34]	纪朋徽, 邢世其, 代大海, 等. 基于逆chirp scaling的合成孔径雷达卷积欺骗干扰方法[J]. 电波科学学报, 2023, 38(6): 1029–1039. doi: 10.12265/j.cjors.2022246. JI Penghui, XING Shiqi, DAI Dahai, et al. A SAR convolutional deceptive jamming method based on the inverse chirp scaling[J]. Chinese Journal of Radio Science, 2023, 38(6): 1029–1039. doi: 10.12265/j.cjors.2022246.
[35]	LIU Yongcai, WANG Wei, PAN Xiaoyi, et al. A frequency-domain three-stage algorithm for active deception jamming against synthetic aperture radar[J]. IET Radar, Sonar & Navigation, 2014, 8(6): 639–646. doi: 10.1049/iet-rsn.2013.0222.
[36]	ZHANG Tianfang, LI Lei, ZHOU Yang, et al. CAS-ViT: Convolutional additive self-attention vision transformers for efficient mobile applications[J]. IEEE Transactions on Image Processing, 2026, 35: 1899–1909. doi: 10.1109/TIP.2026.3655121.
[37]	LIU Xinyu, PENG Houwen, ZHENG Ningxin, et al. EfficientViT: Memory efficient vision transformer with cascaded group attention[C]. 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Vancouver, Canada, 2023: 14420–14430. doi: 10.1109/CVPR52729.2023.01386.
[38]	PAN Junting, BULAT A, TAN Fuwen, et al. EdgeViTs: Competing light-weight CNNs on mobile devices with vision transformers[C]. 17th European Conference on Computer Vision, Tel Aviv, Israel, 2022: 294–311. doi: 10.1007/978-3-031-20083-0_18.
[39]	GERRY M J, POTTER L C, GUPTA I J, et al. A parametric model for synthetic aperture radar measurements[J]. IEEE Transactions on Antennas and Propagation, 1999, 47(7): 1179–1188. doi: 10.1109/8.785750.
[40]	段佳. SAR/ISAR目标电磁特征提取及应用研究[D]. [博士论文], 西安电子科技大学, 2015. doi: 10.7666/d.Y2954112. DUAN J. Study on electro-magnetic feature extraction of SAR/ISAR and its applications[D]. [Ph.D. dissertation], Xidian University, 2015. doi: 10.7666/d.Y2954112.
[41]	HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, USA, 2016: 770–778. doi: 10.1109/CVPR.2016.90.
[42]	SZEGEDY C, IOFFE S, VANHOUCKE V, et al. Inception-v4, inception-ResNet and the impact of residual connections on learning[C]. The 31st AAAI Conference on Artificial Intelligence, San Francisco, USA, 2017: 4278–4284. doi: 10.1609/aaai.v31i1.11231.
[43]	CHOLLET F. Xception: Deep learning with depthwise separable convolutions[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, USA, 2017: 1800–1807. doi: 10.1109/CVPR.2017.195.
[44]	SU Xiu, HUANG Tao, LI Yanxi, et al. Prioritized architecture sampling with Monto-Carlo tree search[C]. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, USA, 2021: 10963–10972. doi: 10.1109/CVPR46437.2021.01082.
[45]	YOU Shan, HUANG Tao, YANG Mingmin, et al. GreedyNAS: Towards fast one-shot NAS with greedy supernet[C]. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, USA, 2020: 1996–2005. doi: 10.1109/CVPR42600.2020.00207.
[46]	CAI Han, WANG Tianzhe, WU Zhanghao, et al. On-device image classification with proxyless neural architecture search and quantization-aware fine-tuning[C]. 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW), Seoul, Korea (South), 2019: 2509–2513. doi: 10.1109/ICCVW.2019.00307.
[47]	LIU Hanxiao, SIMONYAN K, and YANG Yiming. DARTS: Differentiable architecture search[C]. 7th International Conference on Learning Representations, New Orleans, USA, 2019: 1–14.
[48]	HUANG Tao, LIU Jiaqi, YOU Shan, et al. Active generation for image classification[C]. 18th European Conference on Computer Vision, Milan, Italy, 2024: 270–286. doi: 10.1007/978-3-031-73195-2_16.