基于特征校正的多对抗域适应方法

doi:10.11959/j.issn.1000-0801.2024014

电信科学 ›› 2024, Vol. 40 ›› Issue (1): 71-82.doi: 10.11959/j.issn.1000-0801.2024014

• 研究与开发 • 上一篇

基于特征校正的多对抗域适应方法

张永¹^,², 刘昊双², 章琪², 刘文哲¹

¹ 湖州师范学院信息工程学院，浙江湖州 313000
² 辽宁师范大学计算机与信息技术学院，辽宁大连 116081

修回日期:2024-01-07 出版日期:2024-01-01 发布日期:2024-01-01
作者简介:张永（1975- ），男，博士，湖州师范学院教授、博士生导师，辽宁师范大学计算机与信息技术学院教授，主要研究方向为数据挖掘、智能计算、情感计算
刘昊双（1996- ），女，辽宁师范大学计算机与信息技术学院硕士生，主要研究方向为数据挖掘、机器学习
章琪（1999- ），女，辽宁师范大学计算机与信息技术学院硕士生，主要研究方向为机器学习、智能计算
刘文哲（1983- ），女，博士，湖州师范学院讲师，主要研究方向为多视角学习、深度学习
基金资助:
国家自然科学基金资助项目(61772252);辽宁省教育厅科学研究经费资助项目(LJKZ0965);湖州市科技计划项目(2022GZ08);湖州市科技计划项目(2023ZD2004)

Multi-adversarial domain adaptation method based on feature correction

Yong ZHANG¹^,², Haoshuang LIU², Qi ZHANG², Wenzhe LIU¹

¹ School of Information Engineering, Huzhou University, Huzhou 313000, China
² School of Computer ＆Information Technology, Liaoning Normal University, Dalian 116081, China

Revised:2024-01-07 Online:2024-01-01 Published:2024-01-01
Supported by:
The National Natural Science Foundation of China(61772252);The Scientific Research Foundation of the Education Department of Liaoning Province(LJKZ0965);The Huzhou Science and Technology Plan Project(2022GZ08);The Huzhou Science and Technology Plan Project(2023ZD2004)

摘要/Abstract

摘要：

领域自适应可以通过对齐源域和目标域的分布将有标签的源域信息迁移到没有标签但相关的目标域。然而，现有的大多数方法仅对源域和目标域的低层特征分布进行对齐，无法捕获样本中的细粒度信息。基于此，提出了一种基于特征校正的多对抗域适应方法。该方法在引入注意力机制以突出可迁移区域的基础上，通过部署特征校正模块对齐两个域之间的高级特征分布，进一步缩小域差异。此外，为了避免单个分类器过度拟合其自身的噪声伪标签，还提出了双分类器协同训练，并利用图神经网络特征聚合的特性生成更精准的源域标签。在3个迁移学习基准数据集上的大量实验证明所提方法的有效性。

关键词: 领域自适应, 迁移学习, 对抗网络, 注意力机制

Abstract:

Domain adaptation can transfer labeled source domain information to an unlabeled but related target domain by aligning the distribution of source domain and target domain.However, most existing methods only align the low-level feature distributions of the source and target domains, failing to capture fine-grained information within the samples.To address this limitation, a feature correction-based multi-adversarial domain adaptation method was proposed.An attention mechanism to highlight transferable regions was introduced in this method and a feature correction module was deployed to align the high-level feature distributions between the two domains, further reducing domain discrepancies.Additionally, to prevent individual classifiers from overfitting their own noisy pseudo-labels,dual classifier co-training was proposed and the feature aggregation property of graph neural networks was utilized to generate more accurate source domain labels.Extensive experiments on three benchmark datasets for transfer learning demonstrate the effectiveness of the proposed method.

Key words: domain adaptation, transfer learning, adversarial network, attention mechanism

中图分类号:

TP181

张永, 刘昊双, 章琪, 刘文哲. 基于特征校正的多对抗域适应方法[J]. 电信科学, 2024, 40(1): 71-82.

Yong ZHANG, Haoshuang LIU, Qi ZHANG, Wenzhe LIU. Multi-adversarial domain adaptation method based on feature correction[J]. Telecommunications Science, 2024, 40(1): 71-82.

图/表 7

图1

表1

表2

表3

表4

图2

图3

参考文献 32

[1]	PAN S J , YANG Q . A survey on transfer learning[J]. IEEE Tran sactions on Knowledge and Data Engineering, 2010,22(10): 1345-1359.
[2]	ZHUANG F , QI Z , DUAN K ,et al. A comprehensive survey on transfer learning[J]. Proceedings of the IEEE, 2021,109(1): 43-76.
[3]	LONG M S , CAO Y , WANG J M ,et al. Learning transferable features with deep adaptation networks[C]// Proceedings of the Proceedings of the 32nd International Conference on International Conference on Machine Learning - Volume 37. New York:ACM Press, 2015: 97-105.
[4]	LONG M S , CAO Z J , WANG J M ,et al. Conditional adversarial domain adaptation[C]// Proceedings of the Proceedings of the 32nd International Conference on Neural Information Processing Systems. New York:ACM Press, 2018: 1647-1657.
[5]	KANG G L , JIANG L , YANG Y ,et al. Contrastive adaptation network for unsupervised domain adaptation[C]// Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2019: 4888-4897.
[6]	MA A , LI J J , LU K ,et al. Adversarial entropy optimization for unsupervised domain adaptation[J]. IEEE Transactions on Neural Networks and Learning Systems, 2022,33(11): 6263-6274.
[7]	PEI Z , CAO Z , LONG M ,et al. Multi-adversarial domain adaptation[C]// Proceedings of the AAAI Conference on Artificial Intelligence. New York:ACM Press, 2018: 3934-3941.
[8]	YAN H L , DING Y K , LI P H ,et al. Mind the class weight bias:weighted maximum mean discrepancy for unsupervised domain adaptation[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2017: 945-954.
[9]	田磊, 唐永强, 张文生 . 两阶段领域自适应学习[J]. 模式识别与人工智能, 2019,32(9): 773-784.
	TIAN L , TANG Y Q , ZHANG W S . Two stage domain adaptation learning[J]. Pattern Recognition and Artificial Intelligence, 2019,32(9): 773-784.
[10]	蔡瑞初, 李嘉豪, 郝志峰 . 基于类内最大均值差异的无监督领域自适应算法[J]. 计算机应用研究, 2020,37(8): 2371-2375.
	CAI R C , LI J H , HAO Z F . Unsupervised domain adaptive algorithm with intra-class maximum mean discrepancy[J]. Application Research of Computers, 2020,37(8): 2371-2375.
[11]	SUN B C , FENG J S , SAENKO K . Return of frustratingly easy domain adaptation[C]// Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence. New York:ACM Press, 2016: 2058-2065.
[12]	YOSINSKI J , CLUNE J , BENGIO Y ,et al. How transferable are features in deep neural networks[J]. Advances in Neural Information Processing Systems, 2014:27.
[13]	SUN B C , SAENKO K . Deep CORAL:correlation alignment for deep domain adaptation[C]// European Conference on Computer Vision. Cham:Springer, 2016: 443-450.
[14]	ZHU Y C , ZHUANG F Z , WANG J D ,et al. Deep subdomain adaptation network for image classification[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021,32(4): 1713-1722.
[15]	LI S , LIU C , LIN Q ,et al. Domain conditioned adaptation network[C]// Proceedings of the AAAI Conference on Artificial Intelligence. Washington:AAAI Press, 2020: 11386-11393.
[16]	GANIN Y , USTINOVA E , AJAKAN H ,et al. Domain-adversarial training of neural networks[J]. Journal of Machine Learning Research, 2016,17(1): 2096-2130.
[17]	SAITO K , WATANABE K , USHIKU Y ,et al. Maximum classifier discrepancy for unsupervised domain adaptation[C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2018: 3723-3732.
[18]	WANG F , JIANG M Q , QIAN C ,et al. Residual attention network for image classification[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2017: 6450-6458.
[19]	KANG G , ZHENG L , YAN Y ,et al. Deep adversarial attention alignment for unsupervised domain adaptation:The benefit of target expectation maximization[C]// Proceedings of the European Conference on Computer Vision. New York:ACM Press, 2018: 420-436.
[20]	WANG X , LI L , YE W ,et al. Transferable attention for domain adaptation[C]// Proceedings of the AAAI Conference on Artificial Intelligence. Washington:AAAI Press, 2019: 5345-5352.
[21]	ESTRACH J B , ZAREMBA W , SZLAM A ,et al. Spectral networks and deep locally connected networks on graphs[C]// Proceedings of the 2nd International Conference on Learning Representations.[S.l.:s.n.], 2014.
[22]	LUO Y , WANG Z , HUANG Z ,et al. Progressive graph learning for open-set domain adaptation[C]// Proceedings of the 37th International Conference on Machine Learning.[S.l.:s.n.], 2020: 6468-6478.
	LUO Y D , WANG Z J , HUANG Z ,et al. Progressive graph learning for open-set domain adaptation[C]// Proceedings of the Proceedings of the 37th International Conference on Machine Learning.New York:ACM Press.[S.l:s.n.], 2020: 6468-6478.
[23]	YANG X , DENG C , LIU T L ,et al. Heterogeneous graph attention network for unsupervised multiple-target domain adaptation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022,44(4): 1992-2003.
[24]	SAENKO K , KULIS B , FRITZ M ,et al. Adapting visual category models to new domains[C]// Proceedings of the 11th European Conference on Computer vision:Part IV. New York:ACM Press, 2010: 213-226.
[25]	LONG M , CAO Y , CAO Z ,et al. Transferable representation learning with deep adaptation networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018,41(12): 3071-3085.
[26]	VENKATESWARA H , EUSEBIO J , CHAKRABORTY S ,et al. Deep hashing network for unsupervised domain adaptation[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2017: 5385-5394.
[27]	DENG J , DONG W , SOCHER R ,et al. ImageNet:a large-scale hierarchical image database[C]// Proceedings of the 2009 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2009: 248-255.
[28]	HE K M , ZHANG X Y , REN S Q ,et al. Deep residual learning for image recognition[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2016: 770-778.
[29]	LONG M S , ZHU H , WANG J M ,et al. Deep transfer learning with joint adaptation networks[C]// Proceedings of the Proceedings of the 34th International Conference on Machine LearningVolume 70. New York:ACM Press, 2017: 2208-2217.
[30]	LI M X , ZHAI Y M , LUO Y W ,et al. Enhanced transport distance for unsupervised domain adaptation[C]// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2020: 13933-13941.
[31]	HUANG J X , GUAN D Y , XIAO A R ,et al. Category contrast for unsupervised domain adaptation in visual tasks[C]// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2022: 1193-1204.
[32]	CHEN M , ZHAO S , LIU H ,et al. Adversarial-learned loss for domain adaptation[C]// Proceedings of the AAAI Conference on Artificial Intelligence. Washington:AAAI Press, 2020: 3521-3528.

数据集	子域	样本数/个	类别数
Office-31	Amazon (A)	2 817	31
	DSLR (D)	498	31
	Webcam (W)	795	31
ImageCLEF-DA	Caltech (C)	600	12
	ImageNet (I)	600	12
	Pascal (P)	600	12
Office-Home	Artistic (Ar)	2 427	65
	Clip Art (Cl)	4 365	65
	Product (Pr)	4 439	65
	Real-World (Rw)	4 357	65

方法	A→W	D→W	W→D	A→D	D→A	W→A	AVG
ResNet-50	68.4%	96.7%	99.3%	68.9%	62.5%	60.7%	76.1%
DAN	83.8%	96.8%	99.5%	78.4%	66.7%	62.7%	80.4%
D-CORAL	77.7%	97.6%	99.7%	81.1%	64.6%	64%	80.8%
DANN	82%	96.9%	99.1%	79.7%	68.2%	67.4%	82.2%
JAN	85.4%	97.4%	99.8%	84.7%	68.6%	70%	84.3%
MADA	90%	97.4%	99.6%	87.8%	70.3%	66.4%	85.2%
CDAN	93.1%	98.2%	100%	89.8%	70.1%	68%	86.6%
TADA	94.3%	98.7%	99.8%	91.6%	72.9%	73%	88.4%
DSAN	93.6%	98.3%	100%	90.2%	73.5%	74.8%	88.4%
ETD	92.1%	100%	100%	88%	71%	67.8%	86.2%
CaCo	89.7%	98.4%	100%	91.7%	73.1%	72.8%	87.6%
本文算法	94%	98.7%	100%	92.4%	73.8%	73.6%	88.8%

方法	I→P	P→I	I→C	C→I	C→P	P→C	AVG
ResNet-50	74.8%	83.9%	91.5%	78.0%	65.5%	91.2%	80.8%
DAN	74.5%	82.2%	92.8%	86.3%	69.2%	89.8%	82.5%
DANN	75.0%	86.0%	96.2%	87.0%	74.3%	91.5%	85.0%
MADA	75.0%	87.9%	96.0%	88.8%	75.2%	92.2%	85.9%
DCAN	80.5%	91.2%	95.7%	91.8%	77.2%	93.3%	88.3%
本文算法	78.7%	91.9%	96.5%	92.3%	77.8%	95.0%	88.7%

方法方法	A→C	A→P	A→R	C→A	C→P	C→R	P→A	P→C	P→R	R→A	R→C	R→P	AVG
ResNet-50	34.9%	50.0%	58.0%	37.4%	41.9%	46.2%	38.5%	31.2%	60.4%	63.9%	41.2%	59.9%	47.0%
DAN	43.6%	57.0%	67.9%	45.8%	56.5%	60.4%	44.0%	43.6%	67.7%	63.1%	51.5%	74.3%	56.3%
DANN	45.6%	59.3%	70.1%	47.0%	58.5%	60.9%	46.1%	43.7%	68.5%	63.2%	51.8%	76.8%	57.6%
JAN	45.9%	61.2%	68.9%	50.4%	59.7%	61.0%	45.8%	43.4%	70.3%	63.9%	52.4%	76.8%	58.3%
CDAN	50.7%	70.6%	76.0%	57.6%	70.0%	70.0%	57.4%	50.9%	77.3%	70.9%	56.7%	81.6%	65.8%
TADA	53.1%	72.3%	77.2%	59.1%	71.2%	72.1%	59.7%	53.1%	78.7%	72.4%	60.0%	82.9%	67.7%
DSAN	54.4%	70.8%	75.4%	60.4%	67.8%	68.0%	62.6%	55.9%	78.5%	73.8%	60.6%	83.1%	67.6%
ALDA	53.7%	70.1%	76.4%	60.2%	72.6%	71.5%	56.8%	51.9%	77.1%	70.2%	56.3%	82.1%	66.6%
本文算法	55.6%	71.3%	78.0%	61.8%	71.8%	73.5%	61.5%	54.5%	80.0%	72.8%	60.9%	83.2%	68.7%

基于特征校正的多对抗域适应方法

Multi-adversarial domain adaptation method based on feature correction

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 7

参考文献 32

相关文章 15

Metrics

推荐阅读 0

[1]	马辉, 王瑞琴, 杨帅. 一种渐进式增长条件生成对抗网络模型[J]. 电信科学, 2023, 39(6): 105-113.
[2]	张永, 刘纪奎, 柯文龙. 基于并行可分离卷积和标签平滑正则化的脑电情感识别[J]. 电信科学, 2023, 39(5): 116-128.
[3]	阿克弘, 胡晓东. 基于GAN数据重构的电信用户流失预测方法[J]. 电信科学, 2023, 39(3): 135-142.
[4]	刘昊双, 张永, 曹莹波. 基于K-means聚类的子结构相关适配迁移学习方法[J]. 电信科学, 2023, 39(3): 124-134.
[5]	戴剑锋, 陈星妤, 董黎刚, 蒋献. 融合混合嵌入与关系标签嵌入的三元组联合抽取方法[J]. 电信科学, 2023, 39(2): 132-144.
[6]	杨蓓, 梁鑫, 尹航, 蒋峥, 佘小明. 基于自注意力机制的大规模MIMO信道状态信息特征向量反馈方法[J]. 电信科学, 2023, 39(11): 128-136.
[7]	吴尚, 沈雷, 王李军, 张如栩, 胡鑫. 基于镜像填充谱与LA-ResNet50的超短波卫星信道分类识别算法[J]. 电信科学, 2023, 39(10): 74-84.
[8]	王金强, 孙闽红, 唐向宏, 仇兆炀, 曾德国. 小样本下雷达复合干扰半监督迁移学习识别方法[J]. 电信科学, 2023, 39(10): 15-28.
[9]	马稼明, 潘路平, 张琰琳. 基于Transformer的互联网暗链检测方法[J]. 电信科学, 2022, 38(Z2): 241-247.
[10]	祝文军, 王思宁, 高晓欣, 郑倩. 基于知识流和迁移学习的负荷预测[J]. 电信科学, 2022, 38(5): 114-123.
[11]	宋勇, 严志伟, 秦玉坤, 赵东明, 叶晓舟, 柴园园, 欧阳晔. 基于矩阵分解和注意力多任务学习的客服投诉工单分类[J]. 电信科学, 2022, 38(2): 103-110.
[12]	金楠, 王瑞琴, 陆悦聪. 基于艾宾浩斯遗忘曲线和注意力机制的推荐算法[J]. 电信科学, 2022, 38(10): 89-97.
[13]	孙金杨, 刘柏嵩, 任豪, 钱江波. TAGAN：一种融合细粒度语义特征的学术论文对抗推荐算法[J]. 电信科学, 2021, 37(8): 57-65.
[14]	华郁秀, 李荣鹏, 赵志峰, 吴建军, 张宏纲. 基于生成对抗网络的MIMO信道估计方法[J]. 电信科学, 2021, 37(6): 14-22.
[15]	郑声晟, 殷海兵, 黄晓峰, 章天杰. 基于GAN的无监督域自适应行人重识别[J]. 电信科学, 2021, 37(2): 99-106.