Dynamic Adaptive Cross-Domain Mean Approximation

doi:10.3778/j.issn.1673-9418.2405070

Abstract

Abstract: Cross-domain mean approximation (CDMA) is an efficient measure of distributional differences between domains. It measures the sample distribution difference between domains by calculating the distance from a sample in one domain to the sample mean in another domain, which in turn can facilitate cross-domain migration of knowledge. However, in practical applications, the marginal and conditional distributions of data are unbalanced, and CDMA pursues equally measuring the difference between the marginal and conditional distributions without considering the difference between the two, which leads to its inefficiency in transfer learning. For this reason, this paper firstly improves CDMA by introducing an adaptation factor and designing dynamic CDMA to evaluate the edge distribution error and conditional distribution error between the source and target domains. Secondly, on the basis of dynamic CDMA, a dynamic adaptive cross-domain mean approximation (DA-CDMA) feature extraction algorithm is proposed to extract features that are invariant between domains in order to realize cross-domain migration of knowledge. In addition, in order to reduce the shift of mean value caused by individual bad samples far away from the mean center during the feature extraction process, a mean update mechanism is proposed to update the mean value within the class to increase the stability and accuracy of migration. Finally, classification experiments are conducted on publicly available migration learning datasets to verify the effectiveness of the method.

Key words: distribution difference, cross-domain mean approximation, dynamic adaptation, distribution difference measurement, classification, mean updating

摘要： 跨域均值逼近（CDMA）是一种高效的领域间分布差异度量方法。它通过计算一个域的样本到另一个域样本均值的距离来度量领域间的样本分布差异，进而可以促进知识的跨领域迁移。然而，在实际应用中，数据的边缘分布和条件分布往往是不平衡的，CDMA平等地追求度量边缘分布和条件分布的差异，而不考虑二者之间的差异性，导致其在迁移学习中的效率不高。对CDMA进行改进，引入适应性因子，设计动态CDMA评估源域和目标域之间的边缘分布误差和条件分布误差；在动态CDMA基础上，提出动态自适应跨域均值逼近（DA-CDMA）特征提取算法，来提取领域间不变的特征，以实现知识的跨领域迁移。此外，特征提取过程中为减小个别远离均值中心的不良样本对均值造成的偏移，提出均值更新机制，在类内进行均值更新，增加迁移的稳定性和准确性。在公开的迁移学习数据集上进行分类实验，验证了该方法的有效性。

关键词: 分布差异, 跨域均值逼近, 动态自适应, 分布差异度量, 分类, 均值更新

LI Huimin, MA Jianwei, ZANG Shaofei, SONG Yanbing. Dynamic Adaptive Cross-Domain Mean Approximation[J]. Journal of Frontiers of Computer Science and Technology, 2025, 19(5): 1241-1251.

李慧敏, 马建伟, 臧绍飞, 宋彦兵. 动态自适应跨域均值逼近[J]. 计算机科学与探索, 2025, 19(5): 1241-1251.

References

[1] 边坤, 梁慧. 基于机器学习的图案分类研究进展[J]. 图学学报, 2023, 44(3): 415-426.
BIAN K, LIANG H. Research progress of pattern classification based on machine learning[J]. Journal of Graphics, 2023, 44(3): 415-426.
[2] 郭迎春, 冯放, 阎刚, 等. 基于自适应融合网络的跨域行人重识别方法[J]. 自动化学报, 2022, 48(11): 2744-2756.
GUO Y C, FENG F, YAN G, et al. Cross-domain person re-identification on adaptive fusion network[J]. Acta Automatica Sinica, 2022, 48(11): 2744-2756.
[3] 陈建炜, 杨帆, 赖永炫. 一种基于信息熵迁移的文本检测模型自蒸馏方法[J]. 自动化学报, 2024, 50(11): 2128-2139.
CHEN J W, YANG F, LAI Y X. Self-distillation via entropy transfer for scene text detection[J]. Acta Automatica Sinica, 2024, 50(11): 2128-2139.
[4] 张明星, 张骁雄, 刘姗姗, 等. 利用知识图谱的推荐系统研究综述[J]. 计算机工程与应用, 2023, 59(4): 30-42.
ZHANG M X, ZHANG X X, LIU S S, et al. Review of recommendation systems using knowledge graph[J]. Computer Engineering and Applications, 2023, 59(4): 30-42.
[5] 黎英, 宋佩华. 迁移学习在医学图像分类中的研究进展[J]. 中国图象图形学报, 2022, 27(3): 672-686.
LI Y, SONG P H. Review of transfer learning in medical image classification[J]. Journal of Image and Graphics, 2022, 27(3): 672-686.
[6] 范苍宁, 刘鹏, 肖婷, 等. 深度域适应综述: 一般情况与复杂情况[J]. 自动化学报, 2021, 47(3): 515-548.
FAN C N, LIU P, XIAO T, et al. A review of deep domain adaptation: general situation and complex situation[J]. Acta Automatica Sinica, 2021, 47(3): 515-548.
[7] ZHANG J S, LI X, TIAN J L, et al. A variational local weighted deep sub-domain adaptation network for remaining useful life prediction facing cross-domain condition[J]. Reliability Engineering & System Safety, 2023, 231: 108986.
[8] LIU S W, JIANG H K, WU Z H, et al. Intelligent fault diagnosis of rotating machinery using a multi-source domain adaptation network with adversarial discrepancy matching[J]. Reliability Engineering & System Safety, 2023, 231: 109036.
[9] 赵小强, 蒋红梅. 基于特征和类别对齐的领域适应算法[J]. 控制与决策, 2022, 37(5): 1203-1210.
ZHAO X Q, JIANG H M. Domain adaptation based on feature-level and class-level alignment[J]. Control and Decision, 2022, 37(5): 1203-1210.
[10] NING Y J, PENG J T, LIU Q Y, et al. Contrastive learning based on category matching for domain adaptation in hyperspectral image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 5301814.
[11] DENG Z Y, ZHOU K Y, LI D, et al. Dynamic instance domain adaptation[J]. IEEE Transactions on Image Processing, 2022, 31: 4585-4597.
[12] LU H M, WU J Z, RUAN Y J, et al. A multi-source transfer learning model based on LSTM and domain adaptation for building energy prediction[J]. International Journal of Electrical Power & Energy Systems, 2023, 149: 109024.
[13] WANG R, SONG C Y, GAO M X, et al. Model-data fusion domain adaptation for battery state of health estimation with fewer data and simplified feature extractor[J]. Journal of Energy Storage, 2023, 60: 106686.
[14] XU Y B, ZHANG Y M, YUE T X, et al. Graph-based domain adaptation few-shot learning for hyperspectral image classification[J]. Remote Sensing, 2023, 15(4): 1125.
[15] JIN C. Cross-database facial expression recognition based on hybrid improved unsupervised domain adaptation[J]. Multimedia Tools and Applications, 2023, 82(1): 1105-1129.
[16] PRONZATO L. Performance analysis of greedy algorithms for minimising a maximum mean discrepancy[J]. Statistics and Computing, 2022, 33(1): 14.
[17] DEV K, ASHRAF Z, MUHURI P K, et al. Deep autoencoder based domain adaptation for transfer learning[J]. Multimedia Tools and Applications, 2022, 81(16): 22379-22405.
[18] SHE Q S, CHEN T, FANG F, et al. Improved domain adaptation network based on Wasserstein distance for motor imagery EEG classification[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2023, 31: 1137-1148.
[19] ZANG S F, CHENG Y H, WANG X S, et al. Cross domain mean approximation for unsupervised domain adaptation[J]. IEEE Access, 2020, 8: 139052-139069.
[20] ZANG S F, ZHANG P F, GUO L L, et al. Transfer extreme learning machine with cross domain mean approximation projection[C]//Proceedings of the 2022 12th International Conference on Information Technology in Medicine and Education. Piscataway: IEEE, 2022: 490-496.
[21] WANG J D, CHEN Y Q, FENG W J, et al. Transfer learning with dynamic distribution adaptation[J]. ACM Transactions on Intelligent Systems and Technology, 2020, 11(1): 1-25.
[22] WANG J D, FENG W J, CHEN Y Q, et al. Visual domain adaptation with manifold embedded distribution alignment[C]//Proceedings of the 26th ACM International Conference on Multimedia. New York: ACM, 2018: 402-410.
[23] GONG B Q, SHI Y, SHA F, et al. Geodesic flow kernel for unsupervised domain adaptation[C]//Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2012: 2066-2073.
[24] WANG J D, CHEN Y Q, HAO S J, et al. Balanced distribution adaptation for transfer learning[C]//Proceedings of the 2017 IEEE International Conference on Data Mining. Piscataway: IEEE, 2017: 1129-1134.
[25] PAN S J, TSANG I W, KWOK J T, et al. Domain adaptation via transfer component analysis[J]. IEEE Transactions on Neural Networks, 2011, 22(2): 199-210.
[26] ZHU Y, WU X D, QIANG J P, et al. Representation learning via an integrated autoencoder for unsupervised domain adaptation[J]. Frontiers of Computer Science, 2023, 17(5): 175334.
[27] HUANG L Q, LIU Z G, DEZERT J. Cross-domain pattern classification with distribution adaptation based on evidence theory[J]. IEEE Transactions on Cybernetics, 2023, 53(2): 718-731.
[28] LEKSHMI R, SANODIYA R K, JOSE B R, et al. Kernelized global-local discriminant information preservation for unsupervised domain adaptation[J]. Applied Intelligence, 2023, 53(21): 25412-25434.