融合权重机制和改进SDIM的偏标记分类算法

doi:10.3778/j.issn.1002-8331.2006-0289

摘要/Abstract

摘要：

偏标记学习指示例的唯一真实标记隐藏在一组候选标记之中，其目的为对候选标记进行消歧，最终学习到真实标记。现有方法只是就示例之间的相似性或者差异性进行了单方面考量，因此当示例的候选标记增多时会出现消歧准确率与分类准确率大幅度下降的问题。针对以上问题，提出了融合权重机制和改进SDIM的偏标记分类算法，在原SDIM（Partial Label Learning by Semantic Difference Maximization）算法的基础上，增加了最小化同类别示例之间欧几里德距离的操作，缩小了同类别示例之间的语义差异，将示例的相似性纳入学习范围。同时通过求解相关系数最大化问题计算各示例权重，将权重机制引入同类别示例的消歧学习中，对示例的差异性进行了充分考虑。UCI合成数据集上的实验结果表明，相比传统算法，该文算法的消歧准确率提升了0.211%~12.613%，分类准确率提升了0.287%~25.695%。

关键词: 偏标记学习, SDIM算法, 语义差异, 权重机制

Abstract:

The meaning of partial label learning is that the only true label is hidden in a group of candidate labels, whose purpose is to disambiguate the candidate labels and finally pick up the true label. The existing methods only take unilateral consideration of the similarity or difference between instances, so when the number of candidate labels have a sharpen increase, the accuracy of disambiguation and classification will be drop significantly. In response to the above problems, this paper proposes the fusion weight mechanism and improves SDIM partial label classification algorithm. On the basis of the original SDIM（Partial Label Learning by Semantic Difference Maximization） algorithm, it is added to minimize the Euclidean distance between instances of the same category, the operation is used to minimize the semantic difference between instances of the same category and it takes account the similarity of the instances into learning. At the same time, the weight of each instance is calculated by solving the correlation coefficient maximization problem, and the weight mechanism is introduced into the disambiguation learning of instances of the same category, so the differences are fully considered. The experimental results on the UCI synthetic data set show that compared with the traditional algorithm, the disambiguation accuracy of this algorithm is increased by 0.211%~12.613%, and the classification accuracy is increased by 0.287%~25.695%.

Key words: partial label learning, SDIM algorithm, semantic difference, weight mechanism

张慧婷，谢红薇，周辉，张昊. 融合权重机制和改进SDIM的偏标记分类算法[J]. 计算机工程与应用, 2021, 57(21): 195-202.

ZHANG Huiting, XIE Hongwei, ZHOU Hui, ZHANG Hao. Fusion Weight Mechanism and Improved SDIM Partial Label Classification Algorithm[J]. Computer Engineering and Applications, 2021, 57(21): 195-202.

参考文献

[1] HüLLERMEIER E，BERINGER J.Learning from ambiguously labeled examples[C]//International Symposium on Intelligent Data Analysis.Berlin，Heidelberg：Springer，2005：168-179.
[2] NGUYEN N，CARUANA R.Classification with partial labels[C]//Knowledge Discovery and Data Mining，2008：551-559.
[3] COUR T，SAPP B，TASKAR B，et al.Learning from partial labels[J].Journal of Machine Learning Research，2011：1501-1536.
[4] SONG J，LIU H，GENG F，et al.Weakly-supervised classification of pulmonary nodules based on shape characters[C]//2016 IEEE 14th International Conference on Dependable，Autonomic and Secure Computing，14th International Conference on Pervasive Intelligence and Computing，2nd International Conference on Big Data Intelligence and Computing and Cyber Science and TechnologyCongress（DASC/PiCom/DataCom/CyberSciTech），2016：228-232.
[5] ZENG Z，XIAO S，JIA K，et al.Learning by associating ambiguously labeled images[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2013：708-715.
[6] LUO J，ORABONA F.Learning from candidate labeling sets[C]//Advances in Neural Information Processing Systems，2010：1504-1512.
[7] 姚佳奇，徐正国，燕继坤，等.基于标签语义相似的动态多标签文本分类算法[J].计算机工程与应用，2020，56（19）：94-98.
YAO J Q，XU Z G，YAN J K，et al.Dynamic multi-label text classification algorithm based on label semantic similarity[J].Computer Engineering and Applications，2020，56（19）：94-98.
[8] 丁珩珂，吴子朝，王毅刚.采用多标签传播的三维模型标注方法[J].计算机工程与应用，2019，55（21）：197-205.
DING H K，WU Z C，WANG Y G.3D model labeling method based on multi-label propagation[J].Computer Engineering and Applications，2019，55（21）：197-205.
[9] 周克旻，周蓉，滕婧，等.多标记室内小型无人机定位与姿态估计方法[J].计算机工程与应用，2019，55（22）：231-238.
ZHOU K M，ZHOU R，TENG J，et al.Method of location and attitude estimation for small indoor UAV with multiple markers[J].Computer Engineering and Applications，2019，55（22）：231-238.
[10] 陈鸿昶，谢天，高超，等.候选标记信息感知的偏标记学习算法[J].电子与信息学报，2019，41（10）：2516-2524.
CHEN H C，XIE T，GAO C，et al.Candidate label-aware partial label learning algorithm[J].Journal of Electronics and Information Technology，2019，41（10）：2516-2524.
[11] ZHANG M，YU F，TANG C，et al.Disambiguation-free partial label learning[J].IEEE Transactions on Knowledge and Data Engineering，2017，29（10）：2155-2167.
[12] ZHANG M，ZHOU B，LIU X，et al.Partial label learning via feature-aware disambiguation[C]//Knowledge Discovery and Data Mining，2016：1335-1344.
[13] ZHANG M，YU F.Solving the partial label learning problem：an instance-based approach[C]//International Conference on Artificial Intelligence，2015：4048-4054.
[14] YU F，ZHANG M.Maximum margin partial label learning[J].Machine Learning，2017，106（4）：573-593.
[15] TANG C，ZHANG M.Confidence-rated discriminative partial label learning[C]//National Conference on Artificial Intelligence，2017：2611-2617.
[16] WU X，ZHANG M.Towards enabling binary decomposition for partial label learning[C]//International Joint Conference on Artificial Intelligence，2018：2868-2874.
[17] REN X，HE W，QU M，et al.Label noise reduction in entity typing by heterogeneous partial-label embedding[C]//Knowledge Discovery and Data Mining，2016：1825-1834.
[18] WANG Q，LI Y，ZHOU Z，et al.Partial label learning with unlabeled data[C]//International Joint Conference on Artificial Intelligence，2019：3755-3761.
[19] FENG L，AN B.Partial label learning by semantic difference maximization[C]//International Joint Conference on Artificial Intelligence，2019：2294-2300.