EchoAlign: bridging generative and discriminative learning under noisy labels

Abstract

Noisy labels severely hinder the accuracy and generalization of machine learning models, especially when caused by ambiguous instance features that complicate reliable annotation. Existing approaches, such as transition-matrix-based label correction, struggle to capture complex relationships between instances and noisy labels, limiting their effectiveness in such scenarios. We present EchoAlign, a framework that bridges generative and discriminative learning under noisy labels. Instead of correcting labels, EchoAlign treats noisy labels (\tilde{Y}) as accurate and modifies corresponding instances (X) to align with them. The framework integrates two components: (1) EchoMod employs controllable generative models to adjust instance features while preserving key instance-level structural cues (e.g., shape and edges) and avoiding excessive distortion, and (2) EchoSelect addresses distribution shifts by retaining a reliable subset of original instances, guided by feature similarity between original and modified samples. This generative-discriminative interplay enables robust learning even in highly noisy settings. Experiments on three benchmark datasets show that EchoAlign outperforms state-of-the-art methods in most evaluated settings. Under 30% instance-dependent noise, EchoSelect retains nearly twice as many correctly labeled samples as competing approaches while maintaining 99% selection accuracy, highlighting the robustness and effectiveness of EchoAlign.

Graphical abstract

Keywords

learning from noisy labels / controllable generative models / instance modification / feature alignment / sample selection / robust machine learning

References

[1] Tan M, Le Q V. Efficientnet: rethinking model scaling for convolutional neural networks. In: Proceedings of the 36th International Conference on Machine Learning. 2019, 6105−6114

[2] Dosovitskiy A, Beyer L, Kolesnikov A, Weissenborn D, Zhai X, Unterthiner T, Dehghani M, Minderer M, Heigold G, Gelly S, Uszkoreit J, Houlsby N. An image is worth 16x16 words: Transformers for image recognition at scale. In: Proceedings of the 9th International Conference on Learning Representations. 2021

[3] Stiennon N, Ouyang L, Wu J, Ziegler D M, Lowe R, Voss C, Radford A, Amodei D, Christiano P F. Learning to summarize from human feedback. In: Proceedings of the 34th International Conference on Neural Information Processing Systems. 2020, 253

[4] Chen H, Wang J, Shah A, Tao R, Wei H, Xie X, Sugiyama M, Raj B. Understanding and mitigating the label noise in pre-training on downstream tasks. In: Proceedings of the 12th International Conference on Learning Representations. 2024

[5] Yu X, Liu T, Gong M, Tao D. Learning with biased complementary labels. In: Proceedings of the 15th European Conference on Computer Vision. 2018, 69−85

[6] Li W, Wang L, Li W, Agustsson E, Van Gool L. WebVision database: visual learning and understanding from Web data. 2017, arXiv preprint arXiv: 1708.02862

[7] Welinder P, Branson S, Belongie S, Perona P. The multidimensional wisdom of crowds. In: Proceedings of the 24th International Conference on Neural Information Processing Systems. 2010, 2424−2432

[8] Zhang J, Sheng V S, Li T, Wu X . Improving crowdsourced label quality using noise correction. IEEE Transactions on Neural Networks and Learning Systems, 2018, 29( 5): 1675–1688

[9] Natarajan N, Dhillon I S, Ravikumar P, Tewari A. Learning with noisy labels. In: Proceedings of the 27th International Conference on Neural Information Processing Systems. 2013, 1196−1204

[10] Gu K, Masotto X, Bachani V, Lakshminarayanan B, Nikodem J, Yin D . An instance-dependent simulation framework for learning with label noise. Machine Learning, 2023, 112( 6): 1871–1896

[11] Chen H, Raj B, Xie X, Wang J. . On catastrophic inheritance of large foundation models.. Journal of Data-Centric Machine Learning Research, 2024, 1( 21): 1–33

[12] Menon A K, Van Rooyen B, Natarajan N . Learning from binary labels with instance-dependent noise. Machine Learning, 2018, 107( 8): 1561–1595

[13] Xia X, Liu T, Han B, Wang N, Gong M, Liu H, Niu G, Tao D, Sugiyama M. Part-dependent label noise: towards instance-dependent label noise. In: Proceedings of the 34th International Conference on Neural Information Processing Systems. 2020, 637

[14] Yao Y, Gong M, Du Y, Yu J, Han B, Zhang K, Liu T. Which is better for learning with noisy labels: the semi-supervised method or modeling label noise? In: Proceedings of the 40th International Conference on Machine Learning. 2023, 39660−39673

[15] Bai Y, Han Z, Yang E, Yu J, Han B, Wang D, Liu T. Subclass-dominant label noise: a counterexample for the success of early stopping. In: Proceedings of the 37th International Conference on Neural Information Processing Systems. 2023, 2988

[16] Han B, Yao Q, Yu X, Niu G, Xu M, Hu W, Tsang I W, Sugiyama M. Co-teaching: robust training of deep neural networks with extremely noisy labels. In: Proceedings of the 32nd International Conference on Neural Information Processing Systems. 2018, 8536−8546

[17] Yu X, Han B, Yao J, Niu G, Tsang I, Sugiyama M. How does disagreement help generalization against label corruption? In: Proceedings of the 36th International Conference on Machine Learning. 2019, 7164−7173

[18] Wang X, Wang S, Shi H, Wang J, Mei T. Co-mining: deep face recognition with noisy labels. In: Proceedings of 2019 IEEE/CVF International Conference on Computer Vision. 2019, 9357−9366

[19] Scott C, Blanchard G, Handy G. Classification with asymmetric label noise: consistency and maximal denoising. In: Proceedings of the 26th Annual Conference on Learning Theory. 2013, 489−511

[20] Scott C. A rate of convergence for mixture proportion estimation, with application to learning from noisy labels. In: Proceedings of the 18th International Conference on Artificial Intelligence and Statistics. 2015, 838−846

[21] Goldberger J, Ben-Reuven E. Training deep neural-networks using a noise adaptation layer. In: Proceedings of the 5th International Conference on Learning Representations. 2017

[22] Berthon A, Han B, Niu G, Liu T, Sugiyama M. Confidence scores make instance-dependent label-noise learning possible. In: Proceedings of the 38th International Conference on Machine Learning. 2021, 825−836

[23] Reed S E, Lee H, Anguelov D, Szegedy C, Erhan D, Rabinovich A. Training deep neural networks on noisy labels with bootstrapping. In: Proceedings of the 3rd International Conference on Learning Representations. 2015

[24] Liu T, Tao D . Classification with noisy labels by importance reweighting. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38( 3): 447–461

[25] Xia X, Liu T, Wang N, Han B, Gong C, Niu G, Sugiyama M. Are anchor points really indispensable in label-noise learning? In: Proceedings of the 33rd International Conference on Neural Information Processing Systems. 2019, 614

[26] Cheng J, Liu T, Ramamohanarao K, Tao D. Learning with bounded instance- and label-dependent label noise. In: Proceedings of the 37th International Conference on Machine Learning. 2020, 167

[27] Yao Y, Liu T, Gong M, Han B, Niu G, Zhang K. Causality encourages the identifiability of instance-dependent label noise. In: Li S, Chu Z, eds. Machine Learning for Causal Inference. Cham: Springer, 2023, 247−264

[28] Liu Y, Cheng H, Zhang K. Identifiability of label noise transition matrix. In: Proceedings of the 40th International Conference on Machine Learning. 2023, 887

[29] Neuberg L G . Causality: models, reasoning, and inference, by Judea pearl, Cambridge University Press, 2000. Econometric Theory, 2003, 19( 4): 675–685

[30] Peters J, Janzing D, Schölkopf B. Elements of Causal Inference: Foundations and Learning Algorithms. Cambridge: The MIT Press, 2017

[31] Yao Y, Liu T, Gong M, Han B, Niu G, Zhang K. Instance-dependent label-noise learning under structural causal model. In: Proceedings of the 35th International Conference on Neural Information Processing Systems. 2021, 337

[32] Mirza M W, Siddiq A, Khan I R . A comparative study of medical image enhancement algorithms and quality assessment metrics on COVID-19 CT images. Signal, Image and Video Processing, 2023, 17( 4): 915–924

[33] Han Z, Gui X J, Sun H, Yin Y, Li S . Towards accurate and robust domain adaptation under multiple noisy environments. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45( 5): 6460–6479

[34] Han Z, Sun H, Yin Y . Learning transferable parameters for unsupervised domain adaptation. IEEE Transactions on Image Processing, 2022, 31: 6424–6439

[35] Arpit D, Jastrzebski S, Ballas N, Krueger D, Bengio E, Kanwal M S, Maharaj T, Fischer A, Courville A C, Bengio Y, Lacoste-Julien S. A closer look at memorization in deep networks. In: Proceedings of the 34th International Conference on Machine Learning. 2017, 233−242

[36] Wu P, Zheng S, Goswami M, Metaxas D, Chen C. A topological filter for learning with label noise. In: Proceedings of the 34th International Conference on Neural Information Processing Systems. 2020, 1795

[37] Kim T, Ko J, Cho S, Choi J, Yun S Y. FINE samples for learning with noisy labels. In: Proceedings of the 35th International Conference on Neural Information Processing Systems. 2021, 1848

[38] Nguyen D T, Mummadi C K, Ngo T P N, Nguyen T H P, Beggel L, Brox T. SELF: learning to filter noisy labels with self-ensembling. In: Proceedings of the 8th International Conference on Learning Representations. 2020

[39] Liu S, Niles-Weed J, Razavian N, Fernandez-Granda C. Early-learning regularization prevents memorization of noisy labels. In: Proceedings of the 34th International Conference on Neural Information Processing Systems. 2020, 1707

[40] Xia X, Liu T, Han B, Gong C, Wang N, Ge Z, Chang Y. Robust early-learning: hindering the memorization of noisy labels. In: Proceedings of the 9th International Conference on Learning Representations. 2021

[41] Lu Y, Bo Y, He W. Noise attention learning: enhancing noise robustness by gradient scaling. In: Proceedings of the 36th International Conference on Neural Information Processing Systems. 2022, 1683

[42] Bai Y, Yang E, Han B, Yang Y, Li J, Mao Y, Niu G, Liu T. Understanding and improving early stopping for learning with noisy labels. In: Proceedings of the 35th International Conference on Neural Information Processing Systems. 2021, 1867

[43] Tanaka D, Ikami D, Yamasaki T, Aizawa K. Joint optimization framework for learning with noisy labels. In: Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2018, 5552−5560

[44] Li J, Socher R, Hoi S C H. DivideMix: learning with noisy labels as semi-supervised learning. In: Proceedings of the 8th International Conference on Learning Representations. 2020

[45] Du H, Yuan H, Huang Z, Zhao P, Zhou X. Sequential recommendation with diffusion models. 2023, arXiv preprint arXiv: 2304.04541

[46] Wang Y, Liu Z, Yang L, Yu P S. Conditional denoising diffusion for sequential recommendation. In: Proceedings of the 28th Pacific-Asia Conference on Knowledge Discovery and Data Mining. 2024, 156−169

[47] Franceschi J Y, Gartrell M, Dos Santos L, Issenhuth T, de Bézenac E, Chen M, Rakotomamonjy A. Unifying GANs and score-based diffusion as generative particle models. In: Proceedings of the 37th International Conference on Neural Information Processing Systems. 2023, 2610

[48] Zhang L, Rao A, Agrawala M. Adding conditional control to text-to-image diffusion models. In: Proceedings of 2023 IEEE/CVF International Conference on Computer Vision. 2023, 3813−3824

[49] Kingma D P, Salimans T, Poole B, Ho J. Variational diffusion models. In: Proceedings of the 35th International Conference on Neural Information Processing Systems. 2021, 1660

[50] Zhuang Y, Yu Y, Kong L, Chen X, Zhang C. DyGen: learning from noisy labels via dynamics-enhanced generative modeling. In: Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 2023, 3674−3686

[51] Chen J, Zhang R, Yu T, Sharma R, Xu Z, Sun T, Chen C. Label-retrieval-augmented diffusion models for learning from noisy labels. In: Proceedings of the 37th International Conference on Neural Information Processing Systems. 2023, 2903

[52] Chen T, Liu Y, Wang Z, Yuan J, You Q, Yang H, Zhou M. Improving in-context learning in diffusion models with visual context-modulated prompts. 2023, arXiv preprint arXiv: 2312.01408

[53] Bose A J, Monti R P, Grover A. . Controllable generative modeling via causal reasoning.. Transactions on Machine Learning Research, 2022,

[54] Radford A, Kim J W, Hallacy C, Ramesh A, Goh G, Agarwal S, Sastry G, Askell A, Mishkin P, Clark J, Krueger G, Sutskever I. Learning transferable visual models from natural language supervision. In: Proceedings of the 38th International Conference on Machine Learning. 2021, 8748−8763

[55] Yang S, Yang E, Han B, Liu Y, Xu M, Niu G, Liu T. Estimating instance-dependent Bayes-label transition matrix using a deep neural network. In: Proceedings of the 39th International Conference on Machine Learning. 2022, 25302−25312

[56] Krizhevsky A. Learning multiple layers of features from tiny images. Toronto: University of Toronto, Dissertation, 2009

[57] Wei J, Zhu Z, Cheng H, Liu T, Niu G, Liu Y. Learning with noisy labels revisited: a study using real-world human annotations. In: Proceedings of the 10th International Conference on Learning Representations. 2022

[58] Xia X, Lu P, Gong C, Han B, Yu J, Yu J . Regularly truncated m-estimators for learning with noisy labels. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024, 46( 5): 3522–3536

[59] Xiao T, Xia T, Yang Y, Huang C, Wang X. Learning from massive noisy labeled data for image classification. In: Proceedings of 2015 IEEE Conference on Computer Vision and Pattern Recognition. 2015, 2691−2699

[60] Ma X, Huang H, Wang Y, Romano S, Erfani S M, Bailey J. Normalized loss functions for deep learning with noisy labels. In: Proceedings of the 37th International Conference on Machine Learning. 2020, 6543−6553

[61] Menon A K, Rawat A S, Reddi S J, Kumar S. Can gradient clipping mitigate label noise? In: Proceedings of the 8th International Conference on Learning Representations. 2020

[62] Zhou X, Liu X, Zhai D, Jiang J, Ji X . Asymmetric loss functions for noise-tolerant learning: theory and applications. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45( 7): 8094–8109

[63] Wei H, Zhuang H, Xie R, Feng L, Niu G, An B, Li Y. Mitigating memorization of noisy labels by clipping the model prediction. In: Proceedings of the 40th International Conference on Machine Learning. 2023, 36868−36886

[64] Han B, Niu G, Yu X, Yao Q, Xu M, Tsang I W, Sugiyama M. SIGUA: forgetting may make learning with noisy labels more robust. In: Proceedings of the 37th International Conference on Machine Learning. 2020, 375

[65] Xia X, Han B, Zhan Y, Yu J, Gong M, Gong C, Liu T. Combating noisy labels with sample selection by mining high-discrepancy examples. In: Proceedings of 2023 IEEE/CVF International Conference on Computer Vision. 2023, 1833−1843

[66] Zhang H, Cissé M, Dauphin Y N, Lopez-Paz D. mixup: beyond empirical risk minimization. In: Proceedings of the 6th International Conference on Learning Representations. 2018

[67] Arazo E, Ortego D, Albert P, O’Connor N E, McGuinness K. Unsupervised label noise modeling and loss correction. In: Proceedings of the 36th International Conference on Machine Learning. 2019, 312−321

[68] He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. In: Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. 2016, 770−778

[69] Yu X, Liu T, Gong M, Batmanghelich K, Tao D. An efficient and provable approach for mixture proportion estimation using linear independence assumption. In: Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2018, 4480−4489

[70] Paul S, Xu Y, von Platen P . Ultra fast ControlNet with Diffusers. 2023,

[71] Zavadski D, Feiden J F, Rother C. CONTROLNet-XS: rethinking the control of text-to-image diffusion models as feedback-control systems. In: Proceedings of the 18th European Conference on Computer Vision. 2024, 343−362

[72] Miyato T, Kataoka T, Koyama M, Yoshida Y. Spectral normalization for generative adversarial networks. In: Proceedings of the 6th International Conference on Learning Representations. 2018

[73] Gulrajani I, Ahmed F, Arjovsky M, Dumoulin V, Courville A. Improved training of wasserstein GANs. In: Proceedings of the 31st International Conference on Neural Information Processing Systems. 2017, 5769−5779

[74] Rifai S, Vincent P, Muller X, Glorot X, Bengio Y. Contractive auto-encoders: explicit invariance during feature extraction. In: Proceedings of the 28th International Conference on Machine Learning. 2011, 833−840

Rights & permissions

Higher Education Press