Label co-occurrence guided nonlinear disambiguation for partial multi-label learning

Abstract

Partial multi-label learning (PML) addresses challenges where each instance is associated with a set of candidate labels that includes both relevant and irrelevant ones. Traditional label disambiguation strategies often overlook the importance of nonlinear subspace structures. The assumption that data points closely adhere to multiple linear subspaces is restrictive and may not hold in certain applications. Linear subspace clustering algorithms frequently struggle with data that lie on multiple nonlinear manifolds, because they focus only on global linear relationships between data points. To address this gap, we propose a novel approach called label co-occurrence guided nonlinear disambiguation for partial multi-label learning (LCND). Specifically, we introduce a label weight-guided kernel low-rank representation to learn an instance affinity matrix in a nonlinear feature space, enabling effective identification of instances with complex nonlinear structures. Meanwhile, we design a weighted Jaccard distance to quantify label relevance by exploiting label frequency and co-occurrence information. By jointly optimizing instance-level and label-level affinity matrices, the proposed method effectively denoises labels under weak supervision. Extensive experimental results demonstrate that LCND significantly outperforms most state-of-the-art PML methods on the vast majority of benchmark datasets and evaluation metrics.

Keywords

Partial multi-label learning / weakly supervised learning / multi-label learning

References

[1] Zhang ML.A review on multi-label learning algorithms.IEEE Trans Knowl Data Eng2014;26:1819-37

[2] Qian W,Ding W,Vong CM.Label correlations-based multi-label feature selection with label enhancement.Eng Appl Artif Intell2024;127:107310

[3] Fu D,Zhang X.Graph relationship-driven label coded mapping and compensation for multi-label textile fiber recognition.Eng Appl Artif Intell2024;133:108484

[4] Zhang P,Hu J.A conditional-weight joint relevance metric for feature relevancy term.Eng Appl Artif Intell2021;106:104481

[5] Han Q,Gao W.Feature relevance and redundancy coefficients for multi-view multi-label feature selection.Inf Sci2024;652:119747

[6] Mu J,Sun W,Wang S.Classifier enhancement based on credible sample selection for partial multi-label learning.Appl Intell2025;55:1006

[7] Xie MK.Partial multi-label learning.Proc AAAI Conf Artif Intell2018;32:

[8] Boutell MR,Shen X.Learning multi-label scene classification.Pattern Recogn2004;37:1757-71

[9] Zhang ML.ML-KNN: a lazy learning approach to multi-label learning.Pattern Recogn2007;40:2038-48

[10] Chen YN. Feature-aware label space dimension reduction for multi-label classification. 2012. https://proceedings.neurips.cc/paper/2012/file/d4c2e4a3297fe25a71d030b67eb83bfc-Paper.pdf. (accessed 2026-06-26)

[11] Zhang J,Tian N.Label noise correction for crowdsourcing using dynamic resampling.Eng Appl Artif Intell2024;133:108439

[12] Sun F,Huang SJ.A deep model for partial multi-label image classification with curriculum-based disambiguation.Mach Intell Res2024;21:801-14

[13] Sun L,Wang T,Jin Y.Partial multi-label learning by low-rank and sparse decomposition.Proc AAAI Conf Artif Intell2019;33:5016-23

[14] Sun L,Liu J,Lang C.Global-local label correlation for partial multi-label learning.IEEE Trans Multimedia2021;24:581-93

[15] Xie MK.Partial multi-label learning with noisy label identification.IEEE Trans Pattern Anal Mach Intell2022;44:3676-87

[16] Liu BQ,Zhang ML.Towards enabling binary decomposition for partial multi-label learning.IEEE Trans Pattern Anal Mach Intell2023;45:13203-17

[17] Zhang P,Hu J.Multi-label feature selection based on high-order label correlation assumption.Entropy2020;22:797

[18] Liu G,Yu Y.Robust subspace segmentation by low-rank representation. In Proceedings of the 27th International Conference on Machine Learning (ICML-10). Omnipress: 2010; pp. 663-70.

[19] Jiang W,Qi H.Robust subspace segmentation via nonconvex low rank representation.Inf Sci2016;340-1:144-58

[20] Xiao S,Xu D.Robust kernel low-rank representation.IEEE Trans Neural Netw Learn Syst2016;27:2268-81

[21] Li Z,Tang J.Robust structured subspace learning for data representation.IEEE Trans Pattern Anal Mach Intell2015;37:2085-98

[22] Tanfous AB,Amor BB.Sparse coding of shape trajectories for facial expression and action recognition.IEEE Trans Pattern Anal Mach Intell2020;42:2594-607

[23] Roweis ST.Nonlinear dimensionality reduction by locally linear embedding.Science2000;290:2323-6

[24] Abdolali M.Beyond linear subspace clustering: a comparative study of nonlinear manifold clustering algorithms.Comput Sci Rev2021;42:100435

[25] Li H,Li X.An adaptive class prototype generation framework for partial label learning.Eng Appl Artif Intell2024;133:108178

[26] Li W,Shao S.Generalized contrastive partial label learning for cross-subject EEG-based emotion recognition.IEEE Trans Instrum Meas2024;73:1-11

[27] Fan J,Wang Z.Partial label learning with competitive learning graph neural network.Eng Appl Artif Intell2022;111:104779

[28] Feng L.Partial label learning by semantic difference maximization. In Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence. 2019; pp. 2294-300.

[29] Liu LP.A conditional multinomial mixture model for superset label learning. In Proceedings of the 26th International Conference on Neural Information Processing Systems. Curran Associates Inc.: 2012; pp. 548-56.

[30] Cour T,Taskar B.Learning from partial labels.J Mach Learn Res2011;12:1501-36https://jmlr.csail.mit.edu/papers/v12/cour11a.html. (accessed 2026-06-26)

[31] Gong C,Tang Y,Yang J.A regularization approach for instance-based superset label learning.IEEE Trans Cybern2018;48:967-78

[32] Gong X,Yuan D.Generalized large margin kNN for partial label learning.IEEE Trans Multimedia2021;24:1055-66

[33] Sun K,Wang J.PP-PLL: probability propagation for partial label learning. In Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2019, Würzburg, Germany, September 16-20, 2019; Springer; 2020; pp. 123-37.

[34] Wang W.Partial label learning with discrimination augmentation. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. Association for Computing Machinery: 2022; pp. 1920-8.

[35] Lyu G,Huang W,Zhang H.Partial label learning via low-rank representation and label propagation.Soft Comput2020;24:5165-76

[36] Wang S,Wang Z,Feng S.Partial label learning with noisy side information.Appl Intell2022;52:12382-96

[37] Gong B,Jing L.Convolutional attention-based contrastive learning for partial label learning.Appl Intell2025;55:868

[38] Li H,Vong CM.Multi-kernel partial label learning using graph contrast disambiguation.Appl Intell2024;54:9760-82

[39] Zhang ML,Tang CZ.Disambiguation-free partial label learning.IEEE Trans Knowl Data Eng2017;29:2155-67

[40] Lyu G,Wang T,Li Y.GM-PLL: Graph matching based partial label learning.IEEE Trans Knowl Data Eng2021;33:521-35

[41] Feng L,Han B. Provably consistent partial-label learning. arXiv 2020, arXiv:2007.08929. Available online: https://doi.org/10.48550/arXiv.2007.08929. (accessed 2026-06-26)

[42] Liu LP. Learnability of the superset label learning problem. In International Conference on Machine Learning. 2014. https://api.semanticscholar.org/CorpusID:17882853. (accessed 2026-06-26)

[43] Xu N,Lv J,Geng X. Progressive purification for instance-dependent partial label learning. In Proceedings of the 40th International Conference on Machine Learning. PMLR: 2023; pp. 38551-65. https://proceedings.mlr.press/v202/xu23l.html. (accessed 2026-06-26)

[44] Zhang ML.Partial multi-label learning via credible label elicitation.IEEE Trans Pattern Anal Mach Intell2021;43:3587-99

[45] Wang H,Zhao Y,Hu T.Discriminative and correlative partial multi-label learning. In Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence. 2019; pp. 3691-7.

[46] Yu G,Domeniconi C,Li Z.Feature-induced partial multi-label learning. In 2018 IEEE international conference on data mining (ICDM), Singapore, Nov 17-20, 2018. IEEE; 2018. pp. 1398-403.

[47] Gong X,Bao W. Understanding partial multi-label learning via mutual information. In 35th Conference on Neural Information Processing Systems (NeurIPS 2021). 2021; pp. 4147-56. https://proceedings.neurips.cc/paper/2021/file/217c0e01c1828e7279051f1b6675745d-Paper.pdf. (accessed 2026-06-26)

[48] Wu JH,Chen QG,Zhang ML.Feature-induced manifold disambiguation for multi-view partial multi-label learning. In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. Association for Computing Machinery: 2020; pp. 557-65.

[49] Hao P,Gao W.Partial multi-label feature selection via subspace optimization.Inf Sci2023;648:119556

[50] Hang JY. Partial multi-label learning with probabilistic graphical disambiguation. In 37th Conference on Neural Information Processing Systems (NeurIPS 2023). 2023; pp. 1339-51. https://proceedings.neurips.cc/paper_files/paper/2023/file/04e05ba5cbc36044f6499d1edf15247e-Paper-Conference.pdf. (accessed 2026-06-26)

[51] Wang Z,Han M,Wan B.PML-ED: a method of partial multi-label learning by using encoder-decoder framework and exploring label correlation.Inf Sci2024;661:120165

[52] Li Z,Gu T,Qi F.Partial multi-label learning via K-means graph transformer.Knowl Based Syst2025;325:114017

[53] Qi F,Feng J.Nonlinear characteristic-driven partial multi-label learning. In Advances in Knowledge Discovery and Data Mining. Springer Nature Singapore: 2026; pp. 442-54.

[54] Yang J,Win KT,Yang Z.Low-rank and sparse representation based learning for cancer survivability prediction.Inf Sci2022;582:573-92

[55] Fan J,Wang Z.Addressing label ambiguity imbalance in candidate labels: measures and disambiguation algorithm.Inf Sci2022;612:1-19

[56] Zhang P.Feature relevance term variation for multi-label feature selection.Appl Intell2021;51:5095-110

[57] Gao W,Li Y.Feature redundancy based on interaction information for multi-label feature selection.IEEE Access2020;8:146050-64

[58] Auricchio G,Giudici P.Multivariate Gini-type discrepancies.Math Models Methods Appl Sci2025;35:1267-96

[59] Lin Z,Su Z.Linearized alternating direction method with adaptive penalty for low-rank representation. In Proceedings of the 25th International Conference on Neural Information Processing Systems. Curran Associates Inc.: 2011; pp. 612-20.

[60] Huiskes MJ.The MIR flickr retrieval evaluation. In Proceedings of the 1st ACM International Conference on Multimedia Information Retrieval. Association for Computing Machinery: 2008; pp. 39-43.

[61] Katakis I,Vlahavas I. Multilabel text classification for automated tag suggestion. 2008. https://api.semanticscholar.org/CorpusID:15676013. (accessed 2026-06-26)

[62] Briggs F,Raich R.The 9th annual MLSP competition: new methods for acoustic classification of multiple simultaneous bird species in a noisy environment. In 2013 IEEE international workshop on machine learning for signal processing (MLSP), Southampton, UK, Sep 22-25, 2013. IEEE; 2013. p. 1-8.

[63] Trohidis K,Kalliris G. Multi-label classification of music into emotions. 2008; pp. 325-30. https://www.researchgate.net/publication/288623967_Multilabel_classification_of_music_into_emotions. (accessed 2026-06-26)

[64] Pestian JP,Matykiewicz P. A shared task involving multi-label classification of clinical free text. In Biological, translational, and clinical language processing, Prague, Czech Republic, 2007. Association for Computational Linguistics: 2007; pp. 97-104. https://aclanthology.org/W07-1013/. (accessed 2026-06-26)

[65] Lyu G,Li Y.Noisy label tolerance: a new perspective of partial multi-label learning.Inf Sci2021;543:454-66

[66] Wang H,Lyu G. Deep partial multi-label learning with graph disambiguation. arXiv 2023, arXiv:2305.05882. Available online: https://doi.org/10.48550/arXiv.2305.05882. (accessed 2026-06-26)

[67] Fan J,Lu W.Partial multi-label learning via transformer to discover discriminative label embeddings.Neurocomputing2026;684:133518

[68] Demšar J.Statistical comparisons of classifiers over multiple data sets.J Mach Learn Res2006;7:1-30https://www.jmlr.org/papers/volume7/demsar06a/demsar06a.pdf. (accessed 2026-06-26)