Pith. sign in

REVIEW 4 major objections 4 minor 57 references

AEON: Adaptive Estimation of Instance-Dependent In-Distribution and Out-of-Distribution Label Noise for Robust Learning

T0 review · 4 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read AEON claims to be the first method that jointly estimates instance-dependent in-distribution and out-of-distribution label noise rates during training, without clean validation data, and it reports accuracy gains over existing methods on…

desk verdict A useful empirical method and benchmark, but the 'noise rate estimation' claim is fitted scalars until a clean-label control shows the rates track actual corruption. read the letter →

arxiv 2501.13389 v1 pith:WXG3X4BM submitted 2025-01-23 cs.CV

Arpit Garg , Cuong Nguyen , Rafael Felix , Yuyuan Liu , Thanh-Toan Do , Gustavo Carneiro This is my paper · ORCID
classification cs.CV
keywords noisylabelsinstance-dependentlabelnoiseout-of-distributiondetectionenergy-basedscoringrateestimationsemi-supervisedlearningcontrastiveciFAIR-100benchmark
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper sets out to show that jointly estimating the two kinds of label corruption seen in real image datasets—closed-set noise, where the true class is inside the training label set, and open-set noise, where it is outside—makes robust training work better than simply detecting and re-weighting noisy samples. Its method, AEON, learns two scalar noise-rate parameters during training and converts them, through Gaussian-quantile thresholds, into soft per-sample weights that route each sample among supervised, unsupervised, and contrastive losses. The authors claim this is the first method to estimate instance-dependent in-distribution and out-of-distribution noise rates together without a clean validation set, and they report accuracy gains on CIFAR-100, ciFAIR-100, Clothing1M, mini-WebVision, and WebFG-496 at about 1.2 times the runtime of the most efficient comparison method. They also construct a benchmark that injects instance-dependent noise of both types into ciFAIR-100, and report that method rankings on this benchmark track real-world dataset rankings more closely than older synthetic benchmarks do. The paper acknowledges that its temperature parameters still require manual tuning.

What carries the argument

The load-bearing object is a dual-stream soft mask with Gaussian-quantile adaptive thresholds. The OOD stream uses an energy score $E_i = -T_E \log \sum_{k=1}^C e^{f_\theta^{(k)}(\hat{x}_i)/T_E}$; the ID stream uses the per-sample classification loss $L_i^s = -\hat{y}_i^\top \log(\mathrm{softmax}(f_\theta(\hat{x}_i)))$. The thresholds of Eqs. (4) and (8), $\tau = \Phi^{-1}(1-\hat\eta;\mu,\sigma^2)$, convert the current estimated rates into cut positions under the assumption that each score distribution is Gaussian, and sigmoid functions with widths $\beta_{\mathrm{ood}}$ and $\beta_{\mathrm{id}}$ produce the weights $w_i^{\mathrm{ood}}$ and $w_i^{\mathrm{id}}$. These weights are the only connection between the estimated rates and the loss, so the method's behavior hinges on them.

What would settle it

Record the empirical distributions of energy scores and losses on a corrupted batch with known ID and OOD noise rates and compare the Gaussian quantile $\Phi^{-1}(1-\hat\eta;\mu,\sigma^2)$ with the empirical quantile at that rate; if they disagree by more than a small fraction of samples on the datasets AEON evaluates, the threshold mechanism is miscalibrated. The direct experiment is to replace Eqs. (4) and (8) with empirical batch quantiles and check whether accuracy changes.

Watch

Extended reading notes

Core claim

AEON's central claim is that explicit, jointly estimated noise rates—not just per-sample detection—are what make learning robust to mixed closed- and open-set label noise. The method carries two learnable scalars $\gamma_{\mathrm{id}}$ and $\gamma_{\mathrm{ood}}$, mapped through sigmoids to the rates $\hat{\eta}_{\mathrm{id}}$ and $\hat{\eta}_{\mathrm{ood}}$. Each batch it computes energy scores $E_i$ and supervised losses $L_i^s$, fits their empirical mean and variance, and sets adaptive thresholds $\tau_{\mathrm{ood}} = \Phi^{-1}(1-\hat{\eta}_{\mathrm{ood}}; \mu_E, \sigma_E^2)$ and $\tau_{\mathrm{id}} = \Phi^{-1}(1-\hat{\eta}_{\mathrm{id}}; \mu_L, \sigma_L^2)$. Sigmoid weights built from these thresholds route each sample among a supervised loss, a pseudo-label unsupervised loss, and a contrastive loss, so the estimated rates directly control the training objective. The paper reports that this one-stage estimator reaches the best accuracies it compares against on CIFAR-100, ciFAIR-100, Clothing1M, mini-WebVision, and WebFG-496, with estimated rates near the injected or previously reported rates.

Load-bearing premise

The clean/noisy cutoffs are computed by treating each batch's energy scores and loss values as approximately Gaussian, so if those score distributions are not Gaussian the thresholds will cut at the wrong quantiles and the sample weights will be miscalibrated.

Editorial extensions

If this is right

  • Robust training can be done in one stage: mixing supervised, pseudo-label, and contrastive objectives with learned weights removes the need for a separate clean-versus-noisy assignment stage.
  • The reported gains grow with noise severity: up to roughly 9 accuracy points on the proposed benchmark at 40% closed-set plus 40% open-set noise, and larger gains at high noise rates on synthetic CIFAR-100.
  • The roughly 1.2x runtime overhead over PLS makes joint noise-rate estimation cheap enough for large-scale datasets such as the 1M-image Clothing1M set.
  • Estimated noise rates are informative: they stay near injected rates in synthetic settings and near previously measured rates on real-world data, which supports using them to audit a dataset's corruption level.
  • The new benchmark's rankings match real-world web-noise rankings, giving the community a cheaper testbed for screening noisy-label methods.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Untested by the paper: replacing the inverse-CDF thresholds in Eqs. (4) and (8) with empirical batch quantiles would separate the benefit of explicit rate estimation from the cost of the Gaussian assumption.
  • Because the rates are learned end-to-end, AEON could in principle track noise rates that drift during training, such as as the model memorizes; the paper reports only converged values, so this remains an open extension.
  • The paper's own limitation note says temperature parameters are manually tuned; automating them is future work, so transferring AEON to new datasets may require per-dataset hyperparameter adjustment.
  • The similarity between the new benchmark's rankings and real-world rankings suggests a possible low-cost screening protocol for noisy-label methods, though the paper does not formalize one.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper proposes AEON, a one-stage method for learning image classifiers under mixed in-distribution (ID) and out-of-distribution (OOD) instance-dependent label noise. AEON introduces two learnable scalars (gamma_id, gamma_ood) whose sigmoid transforms are called estimated noise rates (eta_hat_id, eta_hat_ood). These rates set adaptive thresholds for energy scores and losses via the inverse CDF of a Gaussian with empirical mean/variance, producing soft sample weights. The training objective combines supervised loss on likely-clean ID samples, unsupervised loss on likely-noisy ID samples, an energy-based OOD loss, and contrastive losses. The paper also proposes a new benchmark (ciFAIR-100 with Places-IDN OOD noise and part-dependent ID noise) and reports state-of-the-art accuracy on this benchmark, on CIFAR-100 with synthetic mixed noise, and on real-world datasets (Clothing1M, mini-WebVision, WebFG-496) with about 1.2x training overhead over PLS.

Significance. If the noise-rate-estimation claim were validated, AEON would be a meaningful step: it tackles ID and OOD noise jointly in one stage, requires no clean validation set, and is computationally efficient. The new benchmark and the consistent accuracy gains are useful empirical contributions. However, the central claim of 'estimating instance-dependent ID and OOD noise rates' is not supported as stated: the reported eta_hat values are fitted parameters that are not shown to be identifiable from the true noise rates. The paper also provides no proof or empirical demonstration that the learned gamma converges to the true rates, and the self-referential construction of thresholds from the same fitted rates weakens the interpretation. The accuracy results are creditable, but the estimation claim needs to be either rigorously validated (e.g., with a clean-label control) or substantially softened.

major comments (4)
  1. [Section 3.2, Eqs. (1), (5), (9)] The estimated noise rates eta_hat_ood and eta_hat_id are learnable scalars that are optimized only through the training objective in Eq. (9), which contains no term that rewards agreement with the true noise rates. The adaptive thresholds in Eqs. (4) and (8) are then derived from these fitted values, so the entire procedure is self-referential: the partition of a batch into clean/noisy is defined by the estimated rates, and the loss then depends on that partition. There is therefore no mechanism that ties the learned gamma to the actual fraction of corrupted samples; on a perfectly clean dataset, nonzero eta_hat could reduce the loss by downweighting hard examples. The near-matching estimates in Tables 1 and 2 are not evidence of measurement without a control experiment. Please add an experiment on clean-label data (e.g., CIFAR-100 with no injected noise) showing that eta_hat converges to approximately 0, or add an identifiable objective (e.g., a clean validation set or a statistical test on the Gaussian mixture) that forces gamma to track the true rates. Without such a test, the paper should not claim to estimate noise rates.
  2. [Section 3.2, Eqs. (4) and (8)] The thresholds tau_ood and tau_id are set to the inverse cumulative distribution function of a Gaussian with the empirical mean and variance of the batch energy scores or loss values. This assumes that the batch energy and loss distributions are approximately Gaussian, but no empirical evidence or theoretical justification is provided. Under instance-dependent noise, loss distributions are often skewed or multimodal (e.g., a mixture of clean and noisy samples), and energy scores for OOD samples may not be Gaussian in practice. If the distributions deviate from Gaussian, the quantile will not correspond to the top eta_hat fraction of noisy samples, causing miscalibrated sample weights and degraded accuracy. Please provide empirical validation of the Gaussian assumption (e.g., histograms or QQ-plots at several training epochs) or replace the inverse-CDF with a nonparametric quantile estimator that does not rely on distributional shape.
  3. [Section 5.2.3] The sentence "These hyperparameters are selected based on performance on a held-out portion of the training set" conflicts with the paper's claim of not requiring clean validation data. A held-out portion of the noisy training set is not a clean set, and selecting hyperparameters on it may still leak noisy-label information into model selection. More importantly, the reported noise rate estimates in Tables 1 and 2 are compared with the injected rates, but the hyperparameters (T, beta, margins, etc.) were presumably tuned on the same benchmark, so the claimed agreement may reflect tuning rather than estimation. Please clarify how the held-out selection was performed, whether the same protocol was applied to all baselines, and how this affects the interpretation of the reported eta_hat values.
  4. [Section 5.3, Tables 1 and 2] The paper uses Expected Calibration Error (ECE) to claim that the dual-stream detection mechanism and noise rate estimation are reliable ("ECE evaluates the reliability of our dual-stream detection mechanism in Eqs. (3) and (7) and noise rate estimation in Eqs. (1) and (5)"). ECE measures the calibration of the classifier's predicted probabilities on the test set; it does not measure the accuracy of eta_hat as an estimate of the true noise rate. The reported ECE improvements therefore do not validate the noise-rate-estimation claim. Please either remove this interpretation or add a direct evaluation of eta_hat against the injected rates over training (e.g., a plot analogous to Fig. 2 for multiple random seeds and noise configurations, with standard deviations).
minor comments (4)
  1. [Section 3.2, Eqs. (4) and (8)] The notation "Phi^{-1}(1 - eta_hat; mu, sigma^2)" is nonstandard: the inverse CDF of a Gaussian is typically parameterized by mean and standard deviation, not variance. Please clarify or correct the notation.
  2. [Section 5.1.1] The sentence "Following our algorithm to implement the benchmark explained in Section 4" is circular in self-reference; it should refer to Section 4's construction method as "the algorithm in Section 4".
  3. [Section 3.3, Eq. (13)] The margin hyperparameter in Eq. (13) is written as "mood", while the text later refers to "m_out" and "m_in". Please unify the notation for clarity.
  4. [Algorithm 1, line 27] The final loss in Algorithm 1 adds Lcont,sup_i and Lcont,uns_i separately, but Eq. (14) already defines Lcont_i as their sum and Eq. (9) includes Lcont_i. Please double-check that the loss is not double-counted.

Circularity Check

1 steps flagged · score 6.0 of 10

The 'estimated noise rates' are sigmoid-transformed free scalars optimized only by the classification/energy/contrastive objective, so the rate-estimation claim reduces to a fitted parameter renamed as a prediction; the held-out accuracy results remain independently evaluated.

  1. fitted input called prediction [Section 3.2, Eq. (1) and Eqs. (4), (5), (8), (9); Tables 1 and 2 rate columns.]
    "We estimate the OOD noise rate, represented by ˆηood∈ [0, 1], through a learnable parameterγood∈ R using a sigmoid function: ˆηood =σ(γood;Tood) = 1 / 1 + e−γood/Tood, (1)"

    Eq. (1) (and its ID analogue, Eq. (5)) defines the 'estimated noise rate' as a sigmoid of a learnable scalar. Eq. (9) optimizes those scalars only through the total classification/energy/contrastive loss, with no term rewarding agreement between the reported rates and the true corruption fractions. The adaptive thresholds in Eqs. (4) and (8) then cut the energy and loss distributions at the quantile implied by these same fitted rates. Therefore the values reported as 'estimated noise rates' in Tables 1-2 and Fig. 2 are free parameters renamed as estimates: nothing ties them to the actual fraction of noisy samples, and on a clean dataset nonzero rates could still lower the loss.

full rationale

The accuracy claims are tested on held-out test sets against external baselines, so those empirical comparisons are independent and keep the paper from being wholly circular. The self-citations [19,20] are used as motivation and background, not as load-bearing uniqueness theorems, and no ansatz is smuggled in through a citation. The circularity is concentrated in the central novelty claim: 'estimating' ID and OOD noise rates is actually fitting two scalars γid and γood through an objective that never measures the true rates. Because the reported noise rates are not identified by any clean-label control or noise-rate supervision, the estimation result reduces to an optimizable parameter renamed as a measurement. That warrants a partial-circularity score of 6 rather than a full 8 or 10.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central method rests on a small number of learned scalars (gamma_id, gamma_ood) and a set of hand-tuned temperatures. The Gaussian assumption for thresholding is the most fragile domain assumption. No invented entities are introduced.

free parameters (3)
  • gamma_ood = learned during training; reported estimates range 0.22-0.63
    Scalar parameter for OOD noise rate, Eq. 1, optimized via the total loss.
  • gamma_id = learned during training; reported estimates range 0.23-0.44
    Scalar parameter for ID noise rate, Eq. 5, optimized via the total loss.
  • Temperature hyperparameters (T_E, T, beta, T_c, gamma_u, margins) = T_E=1, T=10, beta=0.1, T_c=0.07, gamma_u=2, m_id=0.2/0.3, m_ood=0.8/0.9
    Chosen by hand and tuned via sensitivity analysis and held-out validation; affect all detection and weighting mechanisms.
assumptions (4)
  • domain assumption Batch energy scores and loss values are approximately Gaussian distributed.
    Eqs. 4 and 8 use the inverse CDF of a Gaussian with empirical mean and variance to set thresholds. No empirical or theoretical justification is given for this distributional assumption.
  • domain assumption Energy scores are reliable indicators of OOD samples.
    Borrowed from prior work (Liu et al. 2020), used in Eq. 2 to detect open-set noise; presumed to hold in the target datasets.
  • domain assumption Small-loss samples are more likely to be clean.
    Standard small-loss selection prior (e.g., DivideMix), used in Eq. 6 and 7 to identify ID noise; not re-derived.
  • ad hoc to paper Learned gamma parameters converge to the true noise rates under the training objective.
    No theoretical guarantee is given; the paper only checks convergence empirically against injected synthetic noise rates.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AEON: Adaptive Estimation of Instance-Dependent In-Distribution and Out-of-Distribution Label Noise for Robust Learning." pith.science (2026). https://pith.science/paper/WXG3X4BM

@misc{pith2026250113389,
  author       = {Pith},
  title        = {Pith review of: AEON: Adaptive Estimation of Instance-Dependent In-Distribution and Out-of-Distribution Label Noise for Robust Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WXG3X4BM}},
  note         = {Machine review of arXiv:2501.13389}
}
read the original abstract

Robust training with noisy labels is a critical challenge in image classification, offering the potential to reduce reliance on costly clean-label datasets. Real-world datasets often contain a mix of in-distribution (ID) and out-of-distribution (OOD) instance-dependent label noise, a challenge that is rarely addressed simultaneously by existing methods and is further compounded by the lack of comprehensive benchmarking datasets. Furthermore, even though current noisy-label learning approaches attempt to find noisy-label samples during training, these methods do not aim to estimate ID and OOD noise rates to promote their effectiveness in the selection of such noisy-label samples, and they are often represented by inefficient multi-stage learning algorithms. We propose the Adaptive Estimation of Instance-Dependent In-Distribution and Out-of-Distribution Label Noise (AEON) approach to address these research gaps. AEON is an efficient one-stage noisy-label learning methodology that dynamically estimates instance-dependent ID and OOD label noise rates to enhance robustness to complex noise settings. Additionally, we introduce a new benchmark reflecting real-world ID and OOD noise scenarios. Experiments demonstrate that AEON achieves state-of-the-art performance on both synthetic and real-world datasets

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

57 extracted references · 55 canonical work pages

  1. [1]

    Is your noise correction noisy? pls: Robust- ness to label noise with two stage detection

    Paul Albert, Eric Arazo, Tarun Krishna, Noel E O’Connor, and Kevin McGuinness. Is your noise correction noisy? pls: Robust- ness to label noise with two stage detection. In Proceedings of the IEEE/CVF winter con- ference on applications of computer vision , pages 118–127, 2023

    work page 2023

  2. [2]

    Embedding con- trastive unsupervised features to cluster in- and out-of-distribution noise in corrupted 15 image datasets

    Paul Albert, Eric Arazo, Noel E O’Connor, and Kevin McGuinness. Embedding con- trastive unsupervised features to cluster in- and out-of-distribution noise in corrupted 15 image datasets. In European Conference on Computer Vision , pages 402–419. Springer, 2022

    work page 2022

  3. [3]

    Addressing out-of-distribution label noise in webly-labelled data

    Paul Albert, Diego Ortego, Eric Arazo, Noel E O’Connor, and Kevin McGuinness. Addressing out-of-distribution label noise in webly-labelled data. In Proceedings of the IEEE/CVF winter conference on applica- tions of computer vision , pages 392–401, 2022

    work page 2022

  4. [4]

    An accurate detection is not all you need to combat label noise in web-noisy datasets

    Paul Albert, Jack Valmadre, Eric Arazo, Tarun Krishna, Noel E O’Connor, and Kevin McGuinness. An accurate detection is not all you need to combat label noise in web-noisy datasets. In European Conference on Com- puter Vision, pages 55–72. Springer, 2025

    work page 2025

  5. [5]

    Learning from noisy examples

    Dana Angluin and Philip Laird. Learning from noisy examples. Machine Learning , 2(4):343–370, 1988

    work page 1988

  6. [6]

    Unsuper- vised label noise modeling and loss correc- tion

    Eric Arazo, Diego Ortego, Paul Albert, Noel O’Connor, and Kevin McGuinness. Unsuper- vised label noise modeling and loss correc- tion. In International Conference on Machine Learning, pages 312–321. PMLR, 2019

    work page 2019

  7. [7]

    Jastr´ ebski, Nicolas Ballas, David Krueger, Emmanuel Bengio, Maxin- der S

    Devansh Arpit, S. Jastr´ ebski, Nicolas Ballas, David Krueger, Emmanuel Bengio, Maxin- der S. Kanwal, Tegan Maharaj, Asja Fischer, Aaron Courville, Yoshua Bengio, and Simon Lacoste-Julien. A closer look at memo- rization in deep networks. In International Conference on Machine Learning, volume 70, pages 233–242. PMLR, 2017

    work page 2017

  8. [8]

    Do we train on test data? purging cifar of near-duplicates

    Bj¨ orn Barz and Joachim Denzler. Do we train on test data? purging cifar of near-duplicates. Journal of Imaging , 6(6):41, 2020

    work page 2020

Show all 57 references
  1. [9]

    Confidence scores make instance-dependent label-noise learning possible

    Antonin Berthon, Bo Han, Gang Niu, Tongliang Liu, and Masashi Sugiyama. Confidence scores make instance-dependent label-noise learning possible. In International Conference on Machine Learning, pages 825–

  2. [10]

    Better plain ViT base- lines for ImageNet-1k

    Lucas Beyer, Xiaohua Zhai, and Alexan- der Kolesnikov. Better plain ViT base- lines for ImageNet-1k. arXiv preprint arXiv:2205.01580, 2022

    2022 arXiv

  3. [11]

    Instance- dependent label-noise learning with manifold-regularized transition matrix esti- mation

    De Cheng, Tongliang Liu, Yixiong Ning, Nannan Wang, Bo Han, Gang Niu, Xinbo Gao, and Masashi Sugiyama. Instance- dependent label-noise learning with manifold-regularized transition matrix esti- mation. In Conference on Computer Vision and Pattern Recognition, pages 16630–16639, 2022

    2022

  4. [12]

    Learn- ing with bounded instance and label- dependent label noise

    Jiacheng Cheng, Tongliang Liu, Kotagiri Ramamohanarao, and Dacheng Tao. Learn- ing with bounded instance and label- dependent label noise. In International Con- ference on Machine Learning , pages 1789–

  5. [13]

    PropMix: Hard sample filtering and proportional mixup for learning with noisy labels

    Filipe R Cordeiro, Vasileios Belagiannis, Ian Reid, and Gustavo Carneiro. PropMix: Hard sample filtering and proportional mixup for learning with noisy labels. InBritish Machine Vision Conference, 2021

    2021

  6. [14]

    ImageNet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. ImageNet: A large-scale hierarchical image database. In Conference on Computer Vision and Pattern Recognition, pages 248–255. IEEE, 2009

    2009

  7. [15]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020

    2010 arXiv

  8. [16]

    SSR: An efficient and robust framework for learning with unknown label noise

    Chen Feng, Georgios Tzimiropoulos, and Ioannis Patras. SSR: An efficient and robust framework for learning with unknown label noise. In British Machine Vision Conference , page NA. BMVA Press, 2022

    2022

  9. [17]

    Noisebox: Towards more effi- cient and effective learning with noisy labels

    Chen Feng, Georgios Tzimiropoulos, and Ioannis Patras. Noisebox: Towards more effi- cient and effective learning with noisy labels. IEEE Transactions on Circuits and Systems for Video Technology, 2024

    2024

  10. [18]

    Manifold dividemix: A semi-supervised con- trastive learning framework for severe label noise

    Fahimeh Fooladgar, Minh Nguyen Nhat To, Parvin Mousavi, and Purang Abolmaesumi. Manifold dividemix: A semi-supervised con- trastive learning framework for severe label noise. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4012–4021, 2024

    2024

  11. [19]

    Instance-dependent noisy label learning via graphical modelling

    Arpit Garg, Cuong Nguyen, Rafael Felix, Thanh-Toan Do, and Gustavo Carneiro. Instance-dependent noisy label learning via graphical modelling. In Winter Conference on Applications of Computer Vision , pages 2288–2298, January 2023

    2023

  12. [20]

    Instance-dependent noisy-label learning with graphical model based noise-rate estimation, 2024

    Arpit Garg, Cuong Nguyen, Rafael Felix, Thanh-Toan Do, and Gustavo Carneiro. Instance-dependent noisy-label learning with graphical model based noise-rate estimation, 2024

    2024

  13. [21]

    Co-teaching: Robust training of deep neural networks with extremely noisy labels

    Bo Han, Quanming Yao, Xingrui Yu, Gang Niu, Miao Xu, Weihua Hu, Ivor Tsang, and 16 Masashi Sugiyama. Co-teaching: Robust training of deep neural networks with extremely noisy labels. In Advances in Neural Information Processing Systems , volume 31, page NA, 2018

    2018

  14. [22]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Conference on Com- puter Vision and Pattern Recognition , pages 770–778, 2016

    2016

  15. [23]

    Identity mappings in deep residual networks

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Identity mappings in deep residual networks. In European Confer- ence on Computer Vision , pages 630–645. Springer, 2016

    2016

  16. [24]

    Revisiting sample weights based method for noisy-label detection and classification

    Tuan Hoang, Hung Tran, Santu Rana, Sunil Gupta, and Svetha Venkatesh. Revisiting sample weights based method for noisy-label detection and classification. In Proceedings of the Asian Conference on Computer Vision (ACCV), pages 4189–4204, December 2024

    2024

  17. [25]

    Beyond synthetic noise: Deep learning on controlled noisy labels

    Lu Jiang, Di Huang, Mason Liu, and Wei- long Yang. Beyond synthetic noise: Deep learning on controlled noisy labels. In Inter- national Conference on Machine Learning , pages 4804–4815. PMLR, 2020

    2020

  18. [26]

    FINE samples for learn- ing with noisy labels

    Taehyeon Kim, Jongwoo Ko, JinHwan Choi, and Se-Young Yun. FINE samples for learn- ing with noisy labels. In Advances in Neural Information Processing Systems , volume 34, page NA, 2021

    2021

  19. [27]

    Learn- ing multiple layers of features from tiny images

    Alex Krizhevsky and Geoffrey Hinton. Learn- ing multiple layers of features from tiny images. Technical report, University of Toronto, 2009

    2009

  20. [28]

    DivideMix: Learning with noisy labels as semi-supervised learning

    Junnan Li, Richard Socher, and Steven CH Hoi. DivideMix: Learning with noisy labels as semi-supervised learning. In Interna- tional Conference on Learning Representa- tions, page NA, 2020

    2020

  21. [29]

    Learning from noisy data with robust representation learning

    Junnan Li, Caiming Xiong, and Steven CH Hoi. Learning from noisy data with robust representation learning. In International Conference on Computer Vision, pages 9485– 9494, 2021

    2021

  22. [30]

    WebVision Database: Visual learning and understanding from web data

    Wen Li, Limin Wang, Wei Li, Eirikur Agusts- son, and Luc Van Gool. WebVision Database: Visual learning and understanding from web data. CoRR, 2017

    2017

  23. [31]

    Instance-dependent label distribu- tion estimation for learning with label noise

    Zehui Liao, Shishuai Hu, Yutong Xie, and Yong Xia. Instance-dependent label distribu- tion estimation for learning with label noise. International Journal of Computer Vision , pages 1–13, 2024

    2024

  24. [32]

    Genera- tive noisy-label learning by implicit dicrimi- native approximation with partial label prior

    Fengbei Liu, Yuanhong Chen, Chong Wang, Yuyuan Liu, and Gustavo Carneiro. Genera- tive noisy-label learning by implicit dicrimi- native approximation with partial label prior. arXiv preprint arXiv:2308.01184 , 2023

    2023

  25. [33]

    Acpl: Anti-curriculum pseudo-labelling for semi-supervised medical image classification

    Fengbei Liu, Yu Tian, Yuanhong Chen, Yuyuan Liu, Vasileios Belagiannis, and Gustavo Carneiro. Acpl: Anti-curriculum pseudo-labelling for semi-supervised medical image classification. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 2...

    2022

  26. [34]

    Early-learning regularization prevents mem- orization of noisy labels

    Sheng Liu, Jonathan Niles-Weed, Narges Razavian, and Carlos Fernandez-Granda. Early-learning regularization prevents mem- orization of noisy labels. In Advances in Neural Information Processing Systems , vol- ume 33, pages 20331–20342, 2020

    2020

  27. [35]

    Energy-based out-of- distribution detection

    Weitang Liu, Xiaoyun Wang, John Owens, and Yixuan Li. Energy-based out-of- distribution detection. Advances in neural information processing systems , 33:21464– 21475, 2020

    2020

  28. [36]

    Ittakestwo: Leveraging peer rep- resentations for semi-supervised lidar seman- tic segmentation

    Yuyuan Liu, Yuanhong Chen, Hu Wang, Vasileios Belagiannis, Ian Reid, and Gustavo Carneiro. Ittakestwo: Leveraging peer rep- resentations for semi-supervised lidar seman- tic segmentation. In European Conference on Computer Vision , pages 81–99. Springer, 2025

    2025

  29. [37]

    Perturbed and strict mean teachers for semi-supervised semantic segmentation

    Yuyuan Liu, Yu Tian, Yuanhong Chen, Feng- bei Liu, Vasileios Belagiannis, and Gustavo Carneiro. Perturbed and strict mean teachers for semi-supervised semantic segmentation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 4258–4267, 2022

    2022

  30. [38]

    Swin transformer: Hierarchi- cal vision transformer using shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yix- uan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchi- cal vision transformer using shifted windows. In Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision , pages 10012–10022, 2021

    2021

  31. [39]

    A ConvNet for the 2020s

    Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, and 17 Saining Xie. A ConvNet for the 2020s. In Conference on Computer Vision and Pattern Recognition, 2022

    2022

  32. [40]

    Pervasive label errors in test sets destabilize machine learning benchmarks

    Curtis Northcutt, Anish Athalye, and Jonas Mueller. Pervasive label errors in test sets destabilize machine learning benchmarks. In Neural Information Processing Systems (NeurIPS), 2021

    2021

  33. [41]

    EvidentialMix: Learning with combined open-set and closed-set noisy labels

    Ragav Sachdeva, Filipe R Cordeiro, Vasileios Belagiannis, Ian Reid, and Gustavo Carneiro. EvidentialMix: Learning with combined open-set and closed-set noisy labels. In Win- ter Conference on Applications of Computer Vision, pages 3607–3615, 2021

    2021

  34. [42]

    ScanMix: learning from severe label noise via semantic clustering and semi- supervised learning

    Ragav Sachdeva, Filipe Rolim Cordeiro, Vasileios Belagiannis, Ian Reid, and Gus- tavo Carneiro. ScanMix: learning from severe label noise via semantic clustering and semi- supervised learning. Pattern Recognition , 134:109121, 2023

    2023

  35. [43]

    Learning from noisy labels with deep neural networks: A survey

    Hwanjun Song, Minseok Kim, Dongmin Park, Yooju Shin, and Jae-Gil Lee. Learning from noisy labels with deep neural networks: A survey. IEEE Transactions on Neural Networks and Learning Systems , 2022

    2022

  36. [44]

    Webly super- vised fine-grained recognition: Benchmark datasets and an approach

    Zeren Sun, Yazhou Yao, Xiu-Shen Wei, Yong- shun Zhang, Fumin Shen, Jianxin Wu, Jian Zhang, and Heng Tao Shen. Webly super- vised fine-grained recognition: Benchmark datasets and an approach. In Proceedings of the IEEE/CVF international conference on computer vision, pages 1060...

    2021

  37. [45]

    A survey on learning with label noise: Taxonomy, meth- ods, and challenges

    Yongcheng Wang, Chuanxin Liu, Jin Zhang, Jiangchao Ren, and Cheng Deng. A survey on learning with label noise: Taxonomy, meth- ods, and challenges. IEEE Transactions on Pattern Analysis and Machine Intelligence , 45(8):9616–9640, 2023

    2023

  38. [46]

    Open-set label noise can improve robustness against inherent label noise

    Hongxin Wei, Lue Tao, Renchunzi Xie, and Bo An. Open-set label noise can improve robustness against inherent label noise. In Advances in Neural Information Processing Systems, volume 34, 2021

    2021

  39. [47]

    Learning with noisy labels revisited: A study using real-world human annotations

    Jiaheng Wei, Zhaowei Zhu, Hao Cheng, Tongliang Liu, Gang Niu, and Yang Liu. Learning with noisy labels revisited: A study using real-world human annotations. In International Conference on Learning Repre- sentations, page NA, 2022

    2022

  40. [48]

    A topological filter for learning with label noise

    Pengxiang Wu, Songzhu Zheng, Mayank Goswami, Dimitris Metaxas, and Chao Chen. A topological filter for learning with label noise. Advances in Neural Information Pro- cessing Systems, 33:21382–21393, 2020

    2020

  41. [49]

    Part-dependent label noise: Towards instance-dependent label noise

    Xiaobo Xia, Tongliang Liu, Bo Han, Nan- nan Wang, Mingming Gong, Haifeng Liu, Gang Niu, Dacheng Tao, and Masashi Sugiyama. Part-dependent label noise: Towards instance-dependent label noise. In Advances in Neural Information Processing Systems, volume 33, pages 7597–7610, 2020

    2020

  42. [50]

    Learning from massive noisy labeled data for image classification

    Tong Xiao, Tian Xia, Yi Yang, Chang Huang, and Xiaogang Wang. Learning from massive noisy labeled data for image classification. In Conference on Computer Vision and Pattern Recognition, pages 2691–2699, 2015

    2015

  43. [51]

    Searching to exploit memorization effect in learning with noisy labels

    Quanming Yao, Hansi Yang, Bo Han, Gang Niu, and James Tin-Yau Kwok. Searching to exploit memorization effect in learning with noisy labels. In International Conference on Machine Learning , pages 10789–10798. PMLR, 2020

    2020

  44. [52]

    Jo-src: A contrastive approach for combating noisy labels

    Yazhou Yao, Zeren Sun, Chuanyi Zhang, Fumin Shen, Qi Wu, Jian Zhang, and Zhen- min Tang. Jo-src: A contrastive approach for combating noisy labels. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 5192–5201, 2021

    2021

  45. [53]

    Instance-dependent label-noise learning under a structural causal model

    Yu Yao, Tongliang Liu, Mingming Gong, Bo Han, Gang Niu, and Kun Zhang. Instance-dependent label-noise learning under a structural causal model. In Advances in Neural Information Processing Systems , volume 34, page NA, 2021

    2021

  46. [54]

    Vision mamba: Efficient visual representation learn- ing with bidirectional state space model

    Lianghui Zhang, Zhenyu Li, Xuanang Xu, Jiasheng Yu, and Chunyu Yang. Vision mamba: Efficient visual representation learn- ing with bidirectional state space model. In International Conference on Learning Repre- sentations, 2024

    2024

  47. [55]

    Centrality and consis- tency: two-stage clean samples identification for learning with instance-dependent noisy labels

    Ganlong Zhao, Guanbin Li, Yipeng Qin, Feng Liu, and Yizhou Yu. Centrality and consis- tency: two-stage clean samples identification for learning with instance-dependent noisy labels. In European Conference on Computer Vision, pages 21–37. Springer, 2022

    2022

  48. [56]

    Contrast to divide: Self-supervised pre-training for learning with noisy labels

    Evgenii Zheltonozhskii, Chaim Baskin, Avi Mendelson, Alex M Bronstein, and Or Litany. Contrast to divide: Self-supervised pre-training for learning with noisy labels. In Winter Conference on Applications of 18 Computer Vision, pages 1657–1667, 2022

    2022

  49. [57]

    Places: A 10 million image database for scene recogni- tion

    Bolei Zhou, Agata Lapedriza, Aditya Khosla, Aude Oliva, and Antonio Torralba. Places: A 10 million image database for scene recogni- tion. IEEE transactions on pattern analysis and machine intelligence , 40(6):1452–1464, 2017. 19

    2017

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.