Multi-Label Noise Transition Matrix Estimation with Label Correlations: Theory and Algorithm

Noisy multi-label learning has garnered increasing attention due to the challenges posed by collecting large-scale accurate labels, making noisy labels a more practical alternative. Motivated by noisy multi-class learning, the introduction of transition matrices can help model multi-label noise and enable the development of statistically consistent algorithms for noisy multi-label learning. However, estimating multi-label noise transition matrices remains a challenging task, as most existing estimators in noisy multi-class learning rely on anchor points and accurate fitting of noisy class posteriors, which is hard to satisfy in noisy multi-label learning. In this paper, we address this problem by first investigating the identifiability of class-dependent transition matrices in noisy multi-label learning. Building upon the identifiability results, we propose a novel estimator that leverages label correlations without the need for anchor points or precise fitting of noisy class posteriors. Specifically, we first estimate the occurrence probability of two noisy labels to capture noisy label correlations. Subsequently, we employ sample selection techniques to extract information implying clean label correlations, which are then used to estimate the occurrence probability of one noisy label when a certain clean label appears. By exploiting the mismatches in label correlations implied by these occurrence probabilities, we demonstrate that the transition matrix becomes identifiable and can be acquired by solving a bilinear decomposition problem. Theoretically, we establish an estimation error bound for our multi-label transition matrix estimator and derive a generalization error bound for our statistically consistent algorithm. Empirically, we validate the effectiveness of our estimator in estimating multi-label noise transition matrices, leading to excellent classification performance.

[1]  Tongliang Liu,et al.  Trustable Co-Label Learning From Multiple Noisy Annotators , 2022, IEEE Transactions on Multimedia.

[2]  Tongliang Liu,et al.  Selective-Supervised Contrastive Learning with Noisy Labels , 2022, 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[3]  Yang Liu Identifiability of Label Noise Transition Matrix , 2022, ICML.

[4]  T. Ridnik,et al.  ML-Decoder: Scalable and Versatile Classification Head , 2021, IEEE Workshop/Winter Conference on Applications of Computer Vision.

[5]  Tongliang Liu,et al.  Learning with Noisy Labels Revisited: A Study Using Real-World Human Annotations , 2021, ICLR.

[6]  Mingming Gong,et al.  Instance-dependent Label-noise Learning under a Structural Causal Model , 2021, NeurIPS.

[7]  Jianxin Wu,et al.  Residual Attention: A Simple but Effective Method for Multi-Label Recognition , 2021, 2021 IEEE/CVF International Conference on Computer Vision (ICCV).

[8]  Robert Birke,et al.  Multi-Label Gold Asymmetric Loss Correction with Single-Label Regulators , 2021, ArXiv.

[9]  Mingming Gong,et al.  Sample Selection with Uncertainty of Losses for Learning with Noisy Labels , 2021, ICLR.

[10]  Sheng-Jun Huang,et al.  CCMN: A General Framework for Learning With Class-Conditional Multi-Label Noise , 2021, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[11]  Pengxiang Wu,et al.  Learning with Feature-Dependent Label Noise: A Progressive Approach , 2021, ICLR.

[12]  Se-Young Yun,et al.  FINE Samples for Learning with Noisy Labels , 2021, NeurIPS.

[13]  Yang Liu,et al.  Clusterability as an Alternative to Anchor Points When Learning with Noisy Labels , 2021, ICML.

[14]  Masashi Sugiyama,et al.  Provably End-to-end Label-Noise Learning without Anchor Points , 2021, ICML.

[15]  Yang Liu,et al.  A Second-Order Approach to Learning with Instance-Dependent Label Noise , 2020, 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[16]  Hwanjun Song,et al.  Robust Learning by Self-Transition for Handling Noisy Labels , 2020, KDD.

[17]  Yu Qiao,et al.  Attention-Driven Dynamic Graph Convolutional Network for Multi-label Image Recognition , 2020, ECCV.

[18]  Ivor W. Tsang,et al.  The Emerging Trends of Multi-Label Learning , 2020, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[19]  Yang Liu,et al.  Learning with Instance-Dependent Label Noise: A Sample Sieve Approach , 2020, ICLR.

[20]  Emanuel Ben Baruch,et al.  Asymmetric Loss For Multi-Label Classification , 2020, 2021 IEEE/CVF International Conference on Computer Vision (ICCV).

[21]  Hwanjun Song,et al.  Learning From Noisy Labels With Deep Neural Networks: A Survey , 2020, IEEE Transactions on Neural Networks and Learning Systems.

[22]  Bin-Bin Gao,et al.  Learning to Discover Multi-Class Attentional Regions for Multi-Label Image Recognition , 2020, IEEE Transactions on Image Processing.

[23]  Mingming Gong,et al.  Class2Simi: A Noise Reduction Perspective on Learning with Noisy Labels , 2020, ICML.

[24]  Gang Niu,et al.  Dual T: Reducing Estimation Error for Transition Matrix in Label-noise Learning , 2020, NeurIPS.

[25]  Gang Niu,et al.  Parts-dependent Label Noise: Towards Instance-dependent Label Noise , 2020, ArXiv.

[26]  Deyu Meng,et al.  Meta Transition Adaptation for Robust Deep Learning with Noisy Labels , 2020, ArXiv.

[27]  Sheng-Jun Huang,et al.  Partial Multi-Label Learning With Noisy Label Identification , 2020, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[28]  Chunhui Zhang,et al.  Coupled-View Deep Classifier Learning from Multiple Noisy Annotators , 2020, AAAI.

[29]  Junnan Li,et al.  DivideMix: Learning with Noisy Labels as Semi-supervised Learning , 2020, ICLR.

[30]  Gang Niu,et al.  Are Anchor Points Really Indispensable in Label-Noise Learning? , 2019, NeurIPS.

[31]  S. Shan,et al.  Weakly Supervised Image Classification Through Noise Regularization , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[32]  Jae-Gil Lee,et al.  SELFIE: Refurbishing Unclean Samples for Robust Deep Learning , 2019, ICML.

[33]  Noel E. O'Connor,et al.  Unsupervised label noise modeling and loss correction , 2019, ICML.

[34]  Xiu-Shen Wei,et al.  Multi-Label Image Recognition With Graph Convolutional Networks , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[35]  Mert R. Sabuncu,et al.  Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labels , 2018, NeurIPS.

[36]  Sheng-Jun Huang,et al.  Partial Multi-Label Learning , 2018, AAAI.

[37]  Masashi Sugiyama,et al.  Co-teaching: Robust training of deep neural networks with extremely noisy labels , 2018, NeurIPS.

[38]  Wei Liu,et al.  Multi-label Learning with Missing Labels Using Mixed Dependency Graphs , 2018, International Journal of Computer Vision.

[39]  O. Shamir,et al.  Size-Independent Sample Complexity of Neural Networks , 2017, COLT.

[40]  Li Fei-Fei,et al.  MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labels , 2017, ICML.

[41]  Yoshua Bengio,et al.  A Closer Look at Memorization in Deep Networks , 2017, ICML.

[42]  Richard Nock,et al.  Making Deep Neural Networks Robust to Label Noise: A Loss Correction Approach , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[43]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[44]  Cheng Soon Ong,et al.  Learning from Corrupted Binary Labels via Class-Probability Estimation , 2015, ICML.

[45]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[46]  D. Tao,et al.  Classification with Noisy Labels by Importance Reweighting , 2014, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[47]  Michael S. Bernstein,et al.  ImageNet Large Scale Visual Recognition Challenge , 2014, International Journal of Computer Vision.

[48]  Qiang Ji,et al.  Multi-label Learning with Missing Labels , 2014, 2014 22nd International Conference on Pattern Recognition.

[49]  Pietro Perona,et al.  Microsoft COCO: Common Objects in Context , 2014, ECCV.

[50]  Nagarajan Natarajan,et al.  Learning with Noisy Labels , 2013, NIPS.

[51]  S. Boucheron,et al.  Concentration Inequalities: A Nonasymptotic Theory of Independence , 2013 .

[52]  C. Matias,et al.  Identifiability of parameters in latent structure models with many observed variables , 2008, 0809.5032.

[53]  J. Demšar Statistical Comparisons of Classifiers over Multiple Data Sets , 2006, J. Mach. Learn. Res..

[54]  Peter L. Bartlett,et al.  Rademacher and Gaussian Complexities: Risk Bounds and Structural Results , 2003, J. Mach. Learn. Res..

[55]  N. Sidiropoulos,et al.  On the uniqueness of multilinear decomposition of N‐way arrays , 2000 .

[56]  M. Talagrand,et al.  Probability in Banach Spaces: Isoperimetry and Processes , 1991 .

[57]  A E Bostwick,et al.  THE THEORY OF PROBABILITIES. , 1896, Science.

[58]  Ruxin Wang,et al.  LR-SVM+: Learning Using Privileged Information with Noisy Labels , 2022, IEEE Transactions on Multimedia.

[59]  Tongliang Liu,et al.  Estimating Noise Transition Matrix with Label Correlations for Noisy Multi-Label Learning , 2022, NeurIPS.

[60]  Tongliang Liu,et al.  Class-Dependent Label-Noise Learning with Cycle-Consistency Regularization , 2022, NeurIPS.

[61]  Chen Gong,et al.  Robust early-learning: Hindering the memorization of noisy labels , 2021, ICLR.

[62]  L. Gool,et al.  The PASCAL visual object classes challenge 2006 (VOC2006) results , 2006 .

[63]  J. Berge,et al.  On uniqueness in candecomp/parafac , 2002 .

[64]  J. Kruskal Three-way arrays: rank and uniqueness of trilinear decompositions, with application to arithmetic complexity and statistics , 1977 .