The reviewed record of science sign in
Pith

arxiv: 2605.24475 · v1 · pith:ZHEYVXYC · submitted 2026-05-23 · cs.CV · cs.AI· cs.MM

Robust Fuzzy Multi-view Learning under View Conflict

Siyuan Duan , Yuan Sun , Dezhong Peng , Yingke Chen , Xi Peng , Peng Hu This is my paper

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel 2026-06-30 14:12 UTCgrok-4.3pith:ZHEYVXYCrecord.jsonopen to challenge →

classification cs.CV cs.AIcs.MM
keywords multi-view classificationview conflictfuzzy set theoryrobust learninguncertainty estimationtrusted multi-view classificationentropy-based fusion
0
0 comments X

The pith

R-FUML represents network outputs as fuzzy memberships and isolates view conflicts via memory effects to enable reliable fusion under misalignment.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper addresses trusted multi-view classification when views conflict during both training and inference, a setting prior methods largely ignore by assuming perfect alignment. It grounds the solution in fuzzy set theory, treating outputs as fuzzy memberships that express category credibility and applying an entropy measure to fuse views while explicitly tracking both per-view uncertainty and inter-view disagreement. A training procedure first uses neural memory effects to flag conflicting samples, then retrains after penalizing the offending views. Experiments on eight public datasets show consistent gains in robustness and uncertainty calibration over fifteen baselines. If the approach holds, multi-view systems can operate without the strict alignment requirement that currently limits deployment.

Core claim

R-FUML models network outputs as fuzzy memberships to quantify category credibility and uses an entropy-based Robust Multi-view Fusion strategy that accounts for both view-specific uncertainty and inter-view conflicts. It further applies Robust Learning Against VC to isolate conflicting samples by leveraging neural networks' memory effects and then retrains the model after penalizing those conflicting views.

What carries the argument

Robust Multi-view Fusion (RMF) strategy combined with Robust Learning Against VC (RLVC) framework, which together use fuzzy memberships and memory-effect isolation to handle uncertainty and view conflicts.

If this is right

  • Multi-view classifiers can maintain performance when views arrive misaligned at test time.
  • Uncertainty estimates become more trustworthy because fusion explicitly discounts conflicting views.
  • Training no longer overfits to spurious alignments that exist only in the training distribution.
  • The method scales to any number of views without requiring explicit alignment modules.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same memory-effect isolation step could be applied to detect label noise in single-view settings.
  • Fuzzy-membership outputs may transfer directly to calibration techniques in other uncertainty-aware domains.
  • If memory effects prove dataset-dependent, an auxiliary conflict-detection head could replace the reliance on training dynamics.

Load-bearing premise

Neural networks' memory effects can reliably isolate conflicting samples so that penalizing those views improves generalization rather than creating new overfitting or bias.

What would settle it

A controlled test set where samples flagged by memory effects are manually verified as non-conflicting, followed by retraining that shows no robustness gain or a drop in accuracy.

Figures

Figures reproduced from arXiv: 2605.24475 by Dezhong Peng, Peng Hu, Siyuan Duan, Xi Peng, Yingke Chen, Yuan Sun.

Figure 1
Figure 1. Figure 1: Illustration of view conflict (VC) and the motivation [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of our R-FUML. First, view-specific networks [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Overview of the Robust Learning Against VC (RLVC) framework. Stage 1 trains the network to overfitting. Stage 2 [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Test accuracy versus epochs on HW, Fashion, Scene, and LandUse datasets with [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Uncertainty density of clean and conflicting multi-view instances during training: SAEML’s results on the Fashion [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Uncertainty density of clean and conflicting multi-view instances during training: FUML’s results on the Fashion dataset. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: R-FUML’s ability to detect conflicting samples with different numbers of iterations on the HW, Fashion, Scene, and [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
read the original abstract

Trusted multi-view classification aims to deliver reliable fusion for accurate predictions and has recently attracted substantial attention in both academia and industry. However, existing TMVC methods typically assume strict alignment across different views during both training and testing phases, which is often impractical in real-world scenarios. This limitation motivates us to revisit TMVC and extend it to a more challenging setting: how to mitigate the impact of view conflict (VC) during both training and inference. To tackle this setting, existing TMVC methods suffer from three critical limitations: underestimated uncertainty, misleading decisions, and overfitting to VC. To address these issues, this paper proposes a novel Robust Fuzzy Multi-View Learning (R-FUML) framework grounded in Fuzzy Set Theory. Specifically, R-FUML models network outputs as fuzzy memberships to quantify category credibility and uses an entropy-based method for reliable multi-view fusion. To this end, we present a Robust Multi-view Fusion (RMF) strategy that accounts for both view-specific uncertainty and inter-view conflicts, thereby alleviating the adverse impacts of VC on decision-making. To identify and conquer VC during training, we further design a Robust Learning Against VC (RLVC) framework. RLVC isolates conflicting samples by leveraging neural networks' memory effects and then retrains the model by applying a penalty to these conflicting views. Extensive experiments across eight public datasets demonstrate that R-FUML consistently outperforms 15 state-of-the-art baselines in robustness and uncertainty estimation. The code will be released upon acceptance.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The paper proposes Robust Fuzzy Multi-View Learning (R-FUML) to extend trusted multi-view classification to settings with view conflict (VC). It models network outputs as fuzzy memberships, introduces Robust Multi-view Fusion (RMF) that incorporates view-specific uncertainty and inter-view conflicts via entropy, and presents Robust Learning Against VC (RLVC) that isolates conflicting samples via neural network memory effects before applying penalties during retraining. Experiments on eight public datasets are reported to show consistent outperformance over 15 baselines in robustness and uncertainty estimation, with code promised upon acceptance.

Significance. If the central claims hold, the work addresses a practically relevant gap in multi-view learning where strict view alignment cannot be assumed. Grounding the fusion in fuzzy set theory and entropy provides a principled way to quantify credibility and conflict. The explicit plan to release code is a clear strength for reproducibility. The significance is tempered by the need for a precise, reproducible definition of the RLVC isolation step.

major comments (2)
  1. [Abstract] Abstract (RLVC paragraph): The claim that RLVC 'isolates conflicting samples by leveraging neural networks' memory effects' is load-bearing for attributing any performance gains to the handling of VC, yet no equation, algorithm, threshold, memory-effect quantification, or validation metric is supplied. Without these, it is impossible to determine whether the isolation step separates true conflicts from normal variation or introduces new bias, undermining the robustness claim.
  2. [Abstract] Abstract (experimental claim): The statement that R-FUML 'consistently outperforms 15 state-of-the-art baselines' across eight datasets is central to the paper's empirical contribution, but the abstract provides no information on the specific metrics, statistical tests, error bars, or ablation isolating the contribution of RMF versus RLVC. This prevents assessment of whether the reported gains are robust or attributable to the proposed components.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. The comments highlight opportunities to improve the clarity of the abstract, and we address each point below with commitments to revision where appropriate.

read point-by-point responses

  1. Referee: [Abstract] Abstract (RLVC paragraph): The claim that RLVC 'isolates conflicting samples by leveraging neural networks' memory effects' is load-bearing for attributing any performance gains to the handling of VC, yet no equation, algorithm, threshold, memory-effect quantification, or validation metric is supplied. Without these, it is impossible to determine whether the isolation step separates true conflicts from normal variation or introduces new bias, undermining the robustness claim.

    Authors: We agree that the abstract's high-level phrasing leaves the RLVC isolation mechanism underspecified for readers relying solely on the abstract. The full manuscript details the RLVC procedure in Section 3.3, including the memory-effect quantification (based on per-view loss trajectory divergence during early training epochs) and the subsequent penalty application. To address the concern directly in the abstract, we will revise the RLVC sentence to include a concise reference to the memory-effect criterion and penalty term, while preserving the abstract's length constraints. This change will make the isolation step more transparent without altering the underlying method. revision: yes

  2. Referee: [Abstract] Abstract (experimental claim): The statement that R-FUML 'consistently outperforms 15 state-of-the-art baselines' across eight datasets is central to the paper's empirical contribution, but the abstract provides no information on the specific metrics, statistical tests, error bars, or ablation isolating the contribution of RMF versus RLVC. This prevents assessment of whether the reported gains are robust or attributable to the proposed components.

    Authors: The abstract is space-limited and therefore omits granular experimental details that appear in Section 4 (including accuracy, ECE for uncertainty, error bars from 5 random seeds, and component ablations). We will revise the final sentence of the abstract to specify the primary metrics (classification accuracy and uncertainty calibration) and note that gains are supported by statistical comparisons. The revised wording will read along the lines of 'outperforms 15 baselines in accuracy and uncertainty estimation across eight datasets, with ablations confirming the contributions of RMF and RLVC.' This provides the requested context while remaining concise. revision: yes

Circularity Check

0 steps flagged

No circularity: high-level description only, no equations or derivations present

full rationale

The provided abstract and text contain no equations, parameter fittings, self-citations, or derivation steps that could reduce any claimed prediction or result to its inputs by construction. Descriptions of RMF and RLVC are purely narrative without mathematical formulations, uniqueness theorems, or ansatzes that might exhibit self-definitional or fitted-input patterns. The method is presented as grounded in Fuzzy Set Theory at a conceptual level, but no specific reductions or load-bearing self-references appear. This is the expected outcome for a paper whose visible content is limited to prose claims without formal chains.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no free parameters, axioms, or invented entities are explicitly stated or derivable from the provided text.

pith-pipeline@v0.9.1-grok · 5802 in / 1134 out tokens · 30057 ms · 2026-06-30T14:12:33.555750+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

65 extracted references · 4 canonical work pages · 1 internal anchor

  1. [1]

    Deep learning for 3d point clouds: A survey,

    Y . Guo, H. Wang, Q. Hu, H. Liu, L. Liu, and M. Bennamoun, “Deep learning for 3d point clouds: A survey,”IEEE transactions on pattern analysis and machine intelligence, vol. 43, no. 12, pp. 4338–4364, 2020

    2020

  2. [2]

    Robust cross- modal alignment learning for cross-scene spatial reasoning and ground- ing,

    Y . Feng, H. Zhu, D. Peng, X. Peng, X. Song, and P. Hu, “Robust cross- modal alignment learning for cross-scene spatial reasoning and ground- ing,”Advances in Neural Information Processing Systems, vol. 38, pp. 49 055–49 075, 2026

    2026

  3. [3]

    Multimodal fusion on low-quality data: A comprehensive survey,

    Q. Zhang, Y . Wei, Z. Han, H. Fu, X. Peng, Q. Hu, C. Deng, C. Xu, J. Wen, D. Hu,et al., “Multimodal fusion on low-quality data: A comprehensive survey,”Information Fusion, p. 104437, 2026

    2026

  4. [4]

    Real-iad: A real-world multi-view dataset for benchmarking versatile industrial anomaly detection,

    C. Wang, W. Zhu, B.-B. Gao, Z. Gan, J. Zhang, Z. Gu, S. Qian, M. Chen, and L. Ma, “Real-iad: A real-world multi-view dataset for benchmarking versatile industrial anomaly detection,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 22 883–22 892

    2024

  5. [5]

    Carp: Visuomotor policy learning via coarse-to-fine au- toregressive prediction,

    Z. Gong, P. Ding, S. Lyu, S. Huang, M. Sun, W. Zhao, Z. Fan, and D. Wang, “Carp: Visuomotor policy learning via coarse-to-fine au- toregressive prediction,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2025, pp. 13 460–13 470

    2025

  6. [6]

    A deep multi-level attentive network for multimodal sentiment analysis,

    A. Yadav and D. K. Vishwakarma, “A deep multi-level attentive network for multimodal sentiment analysis,”ACM Transactions on Multimedia Computing, Communications and Applications, vol. 19, no. 1, pp. 1–19, 2023

    2023

  7. [7]

    Distilled prompt learning for incomplete multimodal survival prediction,

    Y . Xu, F. Zhou, C. Zhao, Y . Wang, C. Yang, and H. Chen, “Distilled prompt learning for incomplete multimodal survival prediction,” inPro- ceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 5102–5111

    2025

  8. [8]

    Trusted multi-view classi- fication with dynamic evidential fusion,

    Z. Han, C. Zhang, H. Fu, and J. T. Zhou, “Trusted multi-view classi- fication with dynamic evidential fusion,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 2, pp. 2551–2566, 2022

    2022

  9. [9]

    Evidential deep learning to quantify classification uncertainty,

    M. Sensoy, L. Kaplan, and M. Kandemir, “Evidential deep learning to quantify classification uncertainty,”Advances in Neural Information Processing Systems, vol. 31, 2018

    2018

  10. [10]

    Dempster-shafer theory,

    G. Shafer, “Dempster-shafer theory,”Encyclopedia of artificial intelli- gence, vol. 1, pp. 330–331, 1992

    1992

  11. [11]

    Trusted multi-view deep learning with opinion aggregation,

    W. Liu, X. Yue, Y . Chen, and T. Denoeux, “Trusted multi-view deep learning with opinion aggregation,” inProceedings of the AAAI Confer- ence on Artificial Intelligence, vol. 36, no. 7, 2022, pp. 7585–7593

    2022

  12. [12]

    Dynamic evidence decoupling for trusted multi-view learning,

    Y . Liu, L. Liu, C. Xu, X. Song, Z. Guan, and W. Zhao, “Dynamic evidence decoupling for trusted multi-view learning,” inProceedings of the 32nd ACM International Conference on Multimedia, 2024, pp. 7269–7277

    2024

  13. [13]

    Enhancing multi-view classification reli- ability with adaptive rejection,

    W. Liu, Y . Chen, and X. Yue, “Enhancing multi-view classification reli- ability with adaptive rejection,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 39, no. 18, 2025, pp. 18 969–18 977

    2025

  14. [14]

    Exploring and exploiting uncertainty for incomplete multi-view classification,

    M. Xie, Z. Han, C. Zhang, Y . Bai, and Q. Hu, “Exploring and exploiting uncertainty for incomplete multi-view classification,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 19 873–19 882

    2023

  15. [15]

    Trusted multi-view learning with label noise,

    C. Xu, Y . Zhang, Z. Guan, and W. Zhao, “Trusted multi-view learning with label noise,” inProceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, IJCAI-24, K. Larson, Ed. International Joint Conferences on Artificial Intelligence Organization, 8 2024, pp. 5263–5271, main Track

    2024

  16. [16]

    Reliable disentanglement multi-view learning against view adversarial attacks,

    X. Wang, S. Duan, Q. Li, G. Duan, Y . Sun, and D. Peng, “Reliable disentanglement multi-view learning against view adversarial attacks,” arXiv preprint arXiv:2505.04046, 2025

    work page arXiv 2025

  17. [17]

    Noise-robust vision- language pre-training with positive-negative learning,

    Z. Huang, M. Yang, X. Xiao, P. Hu, and X. Peng, “Noise-robust vision- language pre-training with positive-negative learning,”IEEE Transac- tions on Pattern Analysis and Machine Intelligence, 2024

    2024

  18. [18]

    Rethinking noisy video-text retrieval via relation-aware alignment,

    H. Lai, G. Xiong, H. Mai, X. Liu, and T. Zhang, “Rethinking noisy video-text retrieval via relation-aware alignment,” inProceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 9231– 9241

    2025

  19. [19]

    Multi-sensor data fusion based on the belief divergence measure of evidences and the belief entropy,

    F. Xiao, “Multi-sensor data fusion based on the belief divergence measure of evidences and the belief entropy,”Information Fusion, vol. 46, pp. 23–32, 2019

    2019

  20. [20]

    Compound credibility for conflicting evidence combination: An autoencoder-k-means approach,

    Q. Shang, H. Li, Y . Deng, and K. H. Cheong, “Compound credibility for conflicting evidence combination: An autoencoder-k-means approach,” IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 52, no. 9, pp. 5602–5610, 2021

    2021

  21. [21]

    Fuzzy sets,

    L. A. Zadeh, “Fuzzy sets,”Information and Control, vol. 8, no. 3, pp. 338–353, 1965

    1965

  22. [22]

    Reliable representation learning for incomplete multi-view missing multi-label classification,

    C. Liu, J. Wen, Y . Xu, B. Zhang, L. Nie, and M. Zhang, “Reliable representation learning for incomplete multi-view missing multi-label classification,”IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025

    2025

  23. [23]

    Multi- domain feature integration based trusted partial multi-view incomplete multi-label learning,

    J. Wen, J. Zhou, X. Lu, Y . Tian, Z. Zhang, L. Shen, and Y . Xu, “Multi- domain feature integration based trusted partial multi-view incomplete multi-label learning,”IEEE Transactions on Pattern Analysis and Ma- chine Intelligence, 2026

    2026

  24. [24]

    Multi- view clustering via canonical correlation analysis,

    K. Chaudhuri, S. M. Kakade, K. Livescu, and K. Sridharan, “Multi- view clustering via canonical correlation analysis,” inProceedings of the 26th Annual International Conference on Machine Learning, 2009, pp. 129–136

    2009

  25. [25]

    Multi-view canonical correlation anal- ysis,

    J. Rupnik and J. Shawe-Taylor, “Multi-view canonical correlation anal- ysis,” inConference on Data Mining and Data Warehouses (SiKDD 2010), vol. 473, 2010, pp. 1–4

    2010

  26. [26]

    Higcin: Hierarchical graph-based cross inference network for group activity recognition,

    R. Yan, L. Xie, J. Tang, X. Shu, and Q. Tian, “Higcin: Hierarchical graph-based cross inference network for group activity recognition,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 6, pp. 6955–6968, 2020

    2020

  27. [27]

    Predictive dynamic fusion,

    B. Cao, Y . Xia, Y . Ding, C. Zhang, and Q. Hu, “Predictive dynamic fusion,”International Conference on Machine Learning, 2024

    2024

  28. [28]

    Sample-weighted fused graph-based semi- supervised learning on multi-view data,

    J. Bi and F. Dornaika, “Sample-weighted fused graph-based semi- supervised learning on multi-view data,”Information Fusion, vol. 104, p. 102175, 2024

    2024

  29. [29]

    Elip: Enhanced visual-language foundation models for image retrieval,

    G. Zhan, Y . Liu, K. Han, W. Xie, and A. Zisserman, “Elip: Enhanced visual-language foundation models for image retrieval,” in2025 Inter- national Conference on Content-Based Multimedia Indexing (CBMI). IEEE, 2025, pp. 1–8

    2025

  30. [30]

    Trusted multi-view classifica- tion,

    Z. Han, C. Zhang, H. Fu, and J. T. Zhou, “Trusted multi-view classifica- tion,” inInternational Conference on Learning Representations, 2020

    2020

  31. [31]

    Reliable conflictive multi-view learning,

    C. Xu, J. Si, Z. Guan, W. Zhao, Y . Wu, and X. Gao, “Reliable conflictive multi-view learning,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 38, no. 14, 2024, pp. 16 129–16 137

    2024

  32. [32]

    Evidential dissonance measure in robust multi-view classification to resist adversarial attack,

    X. Yue, Z. Dong, Y . Chen, and S. Xie, “Evidential dissonance measure in robust multi-view classification to resist adversarial attack,”Information Fusion, vol. 113, p. 102605, 2025

    2025

  33. [33]

    Uncertainty-aware dual-evidential learning for weakly-supervised temporal action localization,

    M. Chen, J. Gao, and C. Xu, “Uncertainty-aware dual-evidential learning for weakly-supervised temporal action localization,”IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023

    2023

  34. [34]

    A survey on bayesian deep learning,

    H. Wang and D.-Y . Yeung, “A survey on bayesian deep learning,”ACM computing surveys (csur), vol. 53, no. 5, pp. 1–37, 2020

    2020

  35. [35]

    Dropout as a bayesian approximation: Representing model uncertainty in deep learning,

    Y . Gal and Z. Ghahramani, “Dropout as a bayesian approximation: Representing model uncertainty in deep learning,” ininternational conference on machine learning. PMLR, 2016, pp. 1050–1059

    2016

  36. [36]

    Simple and scalable predictive uncertainty estimation using deep ensembles,

    B. Lakshminarayanan, A. Pritzel, and C. Blundell, “Simple and scalable predictive uncertainty estimation using deep ensembles,”Advances in Neural Information Processing Systems, vol. 30, 2017

    2017

  37. [37]

    Greedy policy search: A simple baseline for learnable test-time aug- mentation,

    A. Lyzhov, Y . Molchanova, A. Ashukha, D. Molchanov, and D. Vetrov, “Greedy policy search: A simple baseline for learnable test-time aug- mentation,” inConference on Uncertainty in Artificial Intelligence. PMLR, 2020, pp. 1308–1317

    2020

  38. [38]

    A comprehensive survey on evidential deep learning and its applications,

    J. Gao, M. Chen, L. Xiang, and C. Xu, “A comprehensive survey on evidential deep learning and its applications,”IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025

    2025

  39. [39]

    Deep evidential hashing for trustworthy cross-modal retrieval,

    Y . Li, L. Zhen, Y . Sun, D. Peng, X. Peng, and P. Hu, “Deep evidential hashing for trustworthy cross-modal retrieval,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 39, no. 17, 2025, pp. 18 566–18 574

    2025

  40. [40]

    Learning with noisy correspondence for cross-modal matching,

    Z. Huang, G. Niu, X. Liu, W. Ding, X. Xiao, H. Wu, and X. Peng, “Learning with noisy correspondence for cross-modal matching,”Ad- vances in Neural Information Processing Systems, vol. 34, pp. 29 406– 29 419, 2021. 12

    2021

  41. [41]

    Cross-modal retrieval with partially mismatched pairs,

    P. Hu, Z. Huang, D. Peng, X. Wang, and X. Peng, “Cross-modal retrieval with partially mismatched pairs,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 8, pp. 9595–9610, 2023

    2023

  42. [42]

    Noisy correspondence rectification via asymmetric similarity learning,

    Y . Wang, Y . Wu, Z. Dai, C. Tian, J. Long, and J. Chen, “Noisy correspondence rectification via asymmetric similarity learning,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 39, no. 20, 2025, pp. 21 384–21 392

    2025

  43. [43]

    Noisy- correspondence learning for text-to-image person re-identification,

    Y . Qin, Y . Chen, D. Peng, X. Peng, J. T. Zhou, and P. Hu, “Noisy- correspondence learning for text-to-image person re-identification,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 27 197–27 206

    2024

  44. [44]

    Roll: Robust noisy pseudo-label learning for multi-view clustering with noisy correspon- dence,

    Y . Sun, Y . Li, Z. Ren, G. Duan, D. Peng, and P. Hu, “Roll: Robust noisy pseudo-label learning for multi-view clustering with noisy correspon- dence,” inProceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 30 732–30 741

    2025

  45. [45]

    Noise- tolerant learning for audio-visual action recognition,

    H. Han, Q. Zheng, M. Luo, K. Miao, F. Tian, and Y . Chen, “Noise- tolerant learning for audio-visual action recognition,”IEEE Transactions on Multimedia, vol. 26, pp. 7761–7774, 2024

    2024

  46. [46]

    arXiv preprint arXiv:2401.16702 , year=

    Y . Lin, J. Zhang, Z. Huang, J. Liu, Z. Wen, and X. Peng, “Multi- granularity correspondence learning from long-term noisy videos,”arXiv preprint arXiv:2401.16702, 2024

    work page arXiv 2024

  47. [47]

    Graph matching with bi-level noisy correspondence,

    Y . Lin, M. Yang, J. Yu, P. Hu, C. Zhang, and X. Peng, “Graph matching with bi-level noisy correspondence,” inProceedings of the IEEE/CVF international conference on computer vision, 2023, pp. 23 362–23 371

    2023

  48. [48]

    Robust multi-view clustering with noisy correspondence,

    Y . Sun, Y . Qin, Y . Li, D. Peng, X. Peng, and P. Hu, “Robust multi-view clustering with noisy correspondence,”IEEE Transactions on Knowledge and Data Engineering, 2024

    2024

  49. [49]

    Liu and B

    B. Liu and B. Liu,Uncertainty theory. Springer, 2010

    2010

  50. [50]

    Fuzzy multimodal learning for trusted cross-modal retrieval,

    S. Duan, Y . Sun, D. Peng, Z. Liu, X. Song, and P. Hu, “Fuzzy multimodal learning for trusted cross-modal retrieval,” inProceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 20 747–20 756

    2025

  51. [51]

    DivideMix: Learning with Noisy Labels as Semi-supervised Learning

    J. Li, R. Socher, and S. C. Hoi, “Dividemix: Learning with noisy labels as semi-supervised learning,”arXiv preprint arXiv:2002.07394, 2020

    work page Pith review arXiv 2002

  52. [52]

    Robust object re-identification with coupled noisy labels,

    M. Yang, Z. Huang, and X. Peng, “Robust object re-identification with coupled noisy labels,”International Journal of Computer Vision, vol. 132, no. 7, pp. 2511–2529, 2024

    2024

  53. [53]

    Co-learning: Learning from noisy labels with self-supervision,

    C. Tan, J. Xia, L. Wu, and S. Z. Li, “Co-learning: Learning from noisy labels with self-supervision,” inProceedings of the 29th ACM international conference on multimedia, 2021, pp. 1405–1413

    2021

  54. [54]

    Learning discriminative dynamics with label corruption for noisy label detection,

    S. Kim, D. Lee, S. Kang, S. Chae, S. Jang, and H. Yu, “Learning discriminative dynamics with label corruption for noisy label detection,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 22 477–22 487

    2024

  55. [55]

    O2u-net: A simple noisy label detection approach for deep neural networks,

    J. Huang, L. Qu, R. Jia, and B. Zhao, “O2u-net: A simple noisy label detection approach for deep neural networks,” inProceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 3326– 3334

    2019

  56. [56]

    Deep Learning is Robust to Massive Label Noise

    D. Rolnick, A. Veit, S. Belongie, and N. Shavit, “Deep learning is robust to massive label noise,”arXiv preprint arXiv:1705.10694, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  57. [57]

    Locus: Learning object classes with unsupervised segmentation,

    J. Winn and N. Jojic, “Locus: Learning object classes with unsupervised segmentation,” inTenth IEEE International Conference on Computer Vision (ICCV’05) Volume 1, vol. 1. IEEE, 2005, pp. 756–763

    2005

  58. [58]

    Metaviewer: Towards a unified multi-view representation,

    R. Wang, H. Sun, Y . Ma, X. Xi, and Y . Yin, “Metaviewer: Towards a unified multi-view representation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 11 590–11 599

    2023

  59. [59]

    Bag-of-visual-words and spatial extensions for land-use classification,

    Y . Yang and S. Newsam, “Bag-of-visual-words and spatial extensions for land-use classification,” inProceedings of the 18th SIGSPATIAL Inter- national Conference on Advances in Geographic Information Systems, 2010, pp. 270–279

    2010

  60. [60]

    Dual contrastive prediction for incomplete multi-view representation learning,

    Y . Lin, Y . Gou, X. Liu, J. Bai, J. Lv, and X. Peng, “Dual contrastive prediction for incomplete multi-view representation learning,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 4, pp. 4447–4461, 2022

    2022

  61. [61]

    Provable dynamic fusion for low-quality multimodal data,

    Q. Zhang, H. Wu, C. Zhang, Q. Hu, H. Fu, J. T. Zhou, and X. Peng, “Provable dynamic fusion for low-quality multimodal data,” inInterna- tional Conference on Machine Learning. PMLR, 2023, pp. 41 753– 41 769

    2023

  62. [62]

    Uncertainty-aware multi-view representation learning,

    Y . Geng, Z. Han, C. Zhang, and Q. Hu, “Uncertainty-aware multi-view representation learning,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 35, no. 9, 2021, pp. 7545–7553

    2021

  63. [63]

    Trusted unified feature-neighborhood dynamics for multi- view classification,

    H. Huang, C. Qin, Z. Liu, K. Ma, J. Chen, H. Fang, C. Ban, H. Sun, and Z. He, “Trusted unified feature-neighborhood dynamics for multi- view classification,” inProceedings of the AAAI conference on artificial intelligence, vol. 39, no. 16, 2025, pp. 17 413–17 421

    2025

  64. [64]

    Beyond equal views: Strength-adaptive evidential multi-view learning,

    C. Xu, Z. Wen, J. Zhao, W. Zhao, J. Yu, H. Chen, Z. Guan, and W. Zhao, “Beyond equal views: Strength-adaptive evidential multi-view learning,” inProceedings of the 33rd ACM International Conference on Multimedia, 2025, pp. 1278–1287

    2025

  65. [65]

    Deep fuzzy multi-view learning for reliable classification,

    S. Duan, Y . Sun, D. Peng, G. Duan, X. Peng, and P. Hu, “Deep fuzzy multi-view learning for reliable classification,” inForty-second International Conference on Machine Learning, 2025

    2025