Pith. sign in

REVIEW 4 major objections 6 minor 35 references

RoNFA: Robust Neural Field-based Approach for Few-Shot Image Classification with Noisy Labels

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a two-neural-field few-shot classifier, RoNFA, keeps accuracy near 99% on MiniImageNet even when 60% of support labels are randomly wrong, by building category prototypes through soft K-means and classifying with…

desk verdict The method is a reasonable extension of prototype/kernel ideas, but the near-perfect accuracy almost certainly reflects frozen-ViT pretraining overlap with MiniImageNet/TieredImageNet, so the central empirical claim does not hold as presented. read the letter →

arxiv 2506.03461 v1 pith:RFYKHJPZ submitted 2025-06-03 cs.CV

Nan Xiang , Lifeng Xing , Dequan Jin This is my paper
classification cs.CV
keywords Few-shotlearningNoisylabelsNeuralfieldmodelSoftK-meansclusteringReceptiveScaleadaptationVisionTransformer
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

Few-shot classifiers learn new categories from a handful of labeled images, so even a small number of wrong labels can destroy accuracy. This paper claims that a classifier which constructs category prototypes from noisy support features can stay almost unaffected: with 60% of support-set labels randomly swapped, its 5-way 5-shot accuracy remains 98.33% on MiniImageNet and 90.57% on TieredImageNet, close to its clean-support results and above what strong baselines achieve on clean labels. The mechanism is to place each category neuron in a neural field at a soft K-means center of that category's noisy support features, then classify a query by which category neuron's adaptive receptive field activates. If the claim holds, label noise need not be corrected or relabeled at all in few-shot settings; robust prototypes can absorb it.

What carries the argument

The central object is the Mexican-hat receptive field defined by Eq. (3): $\phi_\sigma(\mathbf{x},\bar{\mathbf{x}}_c)=A e^{-\frac{1}{2}\|\mathbf{x}-\bar{\mathbf{x}}_c\|_2^2/\sigma^2}-B e^{-\frac{1}{2}\|\mathbf{x}-\bar{\mathbf{x}}_c\|_2^2/(3\sigma)^2}$ with $A=1.5$, $B=0.5$. Each category neuron in the category field has one such field centered at the soft K-means prototype $\bar{\mathbf{x}}_c$; a query feature activates neurons only when their field response exceeds a resting level. The scale-adaptation loop shrinks or grows $\sigma$ until exactly one neuron fires, turning the kernel width into an adaptive decision boundary. The soft K-means update in Eq. (1) supplies the prototype centers that the fields lock onto, which is what makes the pipeline insensitive to noisy labels.

What would settle it

Measure, on the same MiniImageNet 5-way 5-shot tasks with 60% symmetric noise, the purity of the soft K-means assignments relative to the true clean labels; if the assignments disagree with the true classes for more than a few percent of support samples, the reported 98.33% accuracy cannot be produced by the clustering mechanism as described. A second check is to rerun the method with a frozen convolutional encoder that yields less separated features and see whether accuracy at 60% noise collapses.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that a two-neural-field classifier—one field holding support features, one holding category neurons with Mexican-hat receptive fields—is enough to make few-shot classification nearly immune to label noise. The model, called RoNFA in the title and VCIM in the body, computes per-class representatives as fixed points of soft K-means initialized at the noisy per-class means, places each category neuron's receptive-field center at that representative, and during prediction adjusts the field scale until exactly one category neuron activates. With a frozen Vision Transformer encoder and no backpropagation, the reported 5-way 5-shot accuracy is 99.11% on MiniImageNet and 94.85% on TieredImageNet at 40% symmetric noise, and 98.33% and 90.57% at 60% symmetric noise, versus clean-support baselines that drop far below those levels. The paper states this as an empirical demonstration of robustness rather than as a theoretical guarantee.

Load-bearing premise

The result rests on soft K-means, initialized at the noisy per-class means, actually finding representatives that match the true categories; in the frozen feature space this requires the classes to be so well separated that cluster recovery succeeds even when, at 60% symmetric noise in a 5-way 5-shot task, each class has on average only two correct labels.

Editorial extensions

If this is right

  • If the claim is right, few-shot systems can tolerate support-set labels that are mostly wrong without any relabeling, loss reweighting, or fine-tuning.
  • The reported numbers imply a noise tolerance ceiling: accuracy drops less than 1% on MiniImageNet and about 5% on TieredImageNet as symmetric noise goes from 0% to 60%.
  • At 60% symmetric noise, the method's accuracy is stated to exceed the clean-support accuracy of the ViT-based ProtoNet and RNNP baselines, which would make noise robustness a property of prototype construction rather than of extra model capacity.
  • Because prediction requires only responses of fixed field neurons rather than iterative training, inference stays parameter-free and cheap, suggesting the method could transfer to settings where per-task fine-tuning is impractical.

Reading between the lines

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

  • Editorial inference: the method's success appears contingent on the frozen encoder's feature separability; a natural stress test is to repeat the protocol with a convolutional encoder or with visually confusable classes, where the cluster-initialization premise would degrade.
  • Editorial inference: the reported 'even surpasses clean-trained baselines' result suggests an upper-bound experiment: measure the clean-label accuracy ceiling of the same two-field classifier; if it is already near 99%, the noise robustness is less surprising than if it is substantially lower.
  • Editorial inference: the scale-adaptation loop can be read as a parameter-free nearest-neighbor rule with an adaptive kernel width; this connection suggests studying the method under distribution shift where no single scale exists, for example mixed-resolution or out-of-distribution queries.
  • Editorial inference: since the method never corrects labels, it may also apply to unsupervised few-shot adaptation where support identities are unknown except for the number of clusters; the paper does not test this.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes RoNFA (also called VCIM), a two-neural-field classifier for few-shot image classification with noisy labels. It uses a frozen ViT feature extractor, soft K-means clustering initialized with per-class support means to obtain class representatives, and a Mexican-hat receptive-field activation with scale adaptation during prediction. Experiments on MiniImageNet and TieredImageNet under symmetric, paired, and outlier label noise report clean accuracies above 99% and 95%, with only small drops at 60% noise, and the authors claim that RoNFA outperforms state-of-the-art FSL methods and even surpasses clean-trained SOTA. The central claim is that the proposed neural-field mechanism provides strong robustness to label noise in the few-shot setting.

Significance. If the reported results were valid, RoNFA would be a notable contribution: a non-backpropagation classifier that maintains roughly 99% accuracy on MiniImageNet at 60% support-label noise would be practically relevant and scientifically interesting. The paper has several strengths: it examines three distinct noise models, includes ablations for soft K-means and scale adaptation, and reports confidence intervals over 600 episodes. However, the evidence base is seriously compromised by the likely overlap between the frozen ViT's pretraining data and the evaluation classes; the paper provides no code, no hyperparameter values, and no analysis of the K-means robustness claim. These issues currently outweigh the strengths, and the manuscript does not establish its central claims.

major comments (4)
  1. [Section 5.1.3 / Tables 1-4] The frozen ViT is almost certainly pretrained on ImageNet, while MiniImageNet [25] and TieredImageNet [18] are ImageNet subsets. The paper does not identify the exact checkpoint, its pretraining corpus, or whether overlapping classes were excluded. The near-perfect accuracies in Tables 1-4 may therefore reflect test-class leakage rather than the proposed classifier's few-shot or noise-robust behavior. Please re-run all experiments with a feature extractor whose pretraining data excludes the benchmark classes (for example, trained only on each benchmark's base split), disclose the checkpoint identity, and report a class-overlap analysis.
  2. [Section 4.1, Eq. (1)] The initial cluster centers μ_c^0 are the means of the (noisy) labeled support samples, so the procedure is not label-insensitive as claimed. The sentence 'The K-means clustering is an unsupervised process insensitive to label noises' is unsupported; at 60% symmetric noise each 5-way 5-shot class has on average only two correct samples, and the initial centers are strongly biased. Please provide a formal condition or an empirical sensitivity analysis (for example, cluster purity versus noise rate, recovery rate of true centers, or comparison with oracle prototypes) to justify the claim that the representatives ̄x_c are correct.
  3. [Abstract / Section 5.2] The claim that noisy-label accuracy surpasses clean-trained SOTA is contradicted by the paper's own tables for symmetric noise. Table 1 reports 98.33% at 60% symmetric noise versus 98.46% for clean ProtoNet VIT and 98.57% for clean RNNP VIT; Table 2 reports 90.57% versus 94.67% and 94.42%. Please restrict the claim to the noise types and settings where it actually holds and revise the abstract accordingly.
  4. [Section 4.3 / Section 5.1.3] The method is not reproducible as reported. No values are given for σ_0, λ, σ_max, σ_min, ε, k_up, h_u, A, B, or the specific ViT variant. The scale-adaptation procedure is under-specified: step 1 sets σ_max = σ_min = 0 but step 2 uses these values, and there is no explicit stopping rule for the case where exactly one neuron is activated. Please provide the hyperparameters, the exact checkpoint, and either working code or a precise pseudocode with convergence behavior.
minor comments (6)
  1. [Section 4.1] The soft-assignment weight formula uses k both as an iteration index and as a category index; the denominator should sum over categories c' = 1, ..., m with distinct indices. Please rewrite with separate indices.
  2. [Section 4.2] The sentence defining the constants says 'A = 1/(3√(2πσ))' where the second constant should be B; the text later says A = 1.5 and B = 0.5, but the derivation should be corrected.
  3. [Section 5.1.1] The dataset description contains a typo: '20 for testing classes for training' should read '20 for testing'.
  4. [Table 1 caption] The caption 'The results in Tables, 1 by [10]' is ungrammatical; it should read 'The results in Tables 1 and 2 are from [10]' or similar.
  5. [Section 5.2] The text says 'accuracy advantage of 1.09% and 3.15% over Vinilla ProtoNet(VIT) with 60% symmetric label swap noise' in the paragraph on outlier noise; it should say 'outlier noise' and 'Vanilla'.
  6. [Section 4.1] The sentence 'we may not relabel some support samples because of their low weights. In this case, we have to abandon them' is unclear: the method does not relabel samples, and it is not explained what abandoning a sample means for the prototype computation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: RoNFA's predictions are computed from support features and unlabeled queries, and all claimed gains are external empirical comparisons.

full rationale

The paper's derivation chain is self-contained rather than circular. The class representatives are obtained by soft K-means clustering on the support set, initialized with per-class means of the (possibly noisy) support labels; no query labels or test statistics enter this construction. The prediction rule then evaluates each query feature against these representatives using a receptive-field response (Eq. 3-4) and adaptively tunes the scale parameter sigma only until exactly one FCR neuron is activated, based on the number of activated neurons, not on the true query label. Thus the adaptive scale is not fitted to the target and the resulting prediction is not equivalent to an input by construction. The paper's core claims are empirical comparisons against external baselines (TraNFS, RNNP, ProtoNet, Matching Networks) on MiniImageNet and TieredImageNet; none of these comparisons reduces to the method's own assumptions. There are no self-citations that carry the argument, and no uniqueness theorem or prior-work ansatz is invoked to forbid alternatives. The assertion in Section 4.1 that K-means 'is an unsupervised process insensitive to label noises' and that the clustering results 'ensure the representatives' is an unproven assumption about feature separability, especially at 60% symmetric noise, but this is a correctness/robustness risk, not circularity. A separate concern is that the frozen ViT backbone may have been pretrained on ImageNet, which overlaps with the MiniImageNet and TieredImageNet test classes; that is a potential leakage threat to the empirical evidence base, but it does not make the derivation circular.

Assumptions & free parameters 5 free parameters · 4 assumptions · 1 invented entities

The central claim depends on unevidenced inputs: hand-chosen hyperparameters (sigma, lambda, h_u, A, B) that are not reported, an unproved assertion that soft K-means corrects noisy labels, and an implicit assumption that the frozen ViT features are nearly perfectly separable. The neural fields are internal constructs without independent evidence.

free parameters (5)
  • sigma_0 (initial receptive field scale) = not reported
    Initial width of the DoG receptive field; directly controls how many FCR neurons fire for a query.
  • lambda, sigma_max, sigma_min (scale adaptation hyperparameters) = not reported
    Used in the Section 4.3 adaptive search; the algorithm's behavior and runtime depend on them.
  • epsilon and k_up (K-means stopping criteria) = not reported
    Stopping condition for the soft K-means iterations in Section 4.1.
  • h_u (resting level) = not reported
    Threshold in Eq. (4) below which a category neuron stays inactive.
  • A, B (DoG amplitudes) = A=1.5, B=0.5
    Set by simplification in Section 4.2, not derived from data.
assumptions (4)
  • domain assumption Soft K-means initialized by per-class means recovers true category representatives under high label noise.
    Section 4.1 claims K-means is 'insensitive to label noises'; this is asserted, not proven, and is the core of the robustness claim.
  • domain assumption The frozen ViT feature space is nearly class-separable for the test categories.
    The 99% clean accuracy on MiniImageNet and the ability to correct 60% noise presuppose features that separate classes almost perfectly; the paper does not show that the ViT was pretrained without overlap on these classes.
  • domain assumption Exactly one activated FCR neuron corresponds to the true category.
    Section 4.3 uses single-neuron activation as the stopping criterion and 'ideal scenario'; no argument links this criterion to ground truth.
  • ad hoc to paper Mexican-hat receptive fields and Hebbian rules provide a valid classification mechanism.
    Sections 4.2 and 4.3 borrow neuroscience-inspired constructs for the classifier; these are architectural choices without independent validation.
invented entities (1)
  • Neural fields FFR and FCR with Mexican-hat receptive fields
    purpose: Represent support-set feature distributions and category activations, and produce predictions by selecting a single activated category neuron.
    These are new model constructs introduced in Section 4.2; they make no falsifiable prediction outside the reported benchmarks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RoNFA: Robust Neural Field-based Approach for Few-Shot Image Classification with Noisy Labels." pith.science (2026). https://pith.science/paper/RFYKHJPZ

@misc{pith2026250603461,
  author       = {Pith},
  title        = {Pith review of: RoNFA: Robust Neural Field-based Approach for Few-Shot Image Classification with Noisy Labels},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RFYKHJPZ}},
  note         = {Machine review of arXiv:2506.03461}
}
read the original abstract

In few-shot learning (FSL), the labeled samples are scarce. Thus, label errors can significantly reduce classification accuracy. Since label errors are inevitable in realistic learning tasks, improving the robustness of the model in the presence of label errors is critical. This paper proposes a new robust neural field-based image approach (RoNFA) for few-shot image classification with noisy labels. RoNFA consists of two neural fields for feature and category representation. They correspond to the feature space and category set. Each neuron in the field for category representation (FCR) has a receptive field (RF) on the field for feature representation (FFR) centered at the representative neuron for its category generated by soft clustering. In the prediction stage, the range of these receptive fields adapts according to the neuronal activation in FCR to ensure prediction accuracy. These learning strategies provide the proposed model with excellent few-shot learning capability and strong robustness against label noises. The experimental results on real-world FSL datasets with three different types of label noise demonstrate that the proposed method significantly outperforms state-of-the-art FSL methods. Its accuracy obtained in the presence of noisy labels even surpasses the results obtained by state-of-the-art FSL methods trained on clean support sets, indicating its strong robustness against noisy labels.

Figures

Figures reproduced from arXiv: 2506.03461 by the authors.

Figure 1
Figure 1. (a) The model extracts features from the input image through the feature extractor. (b) Correcting the class distribution of label-noise samples using soft K-means clustering. (c) Calculating class representative samples after clustering using a weighted mean. (d) Classifying query set samples based on visual cognitive mechanisms and using scale adaptation. When a neuron in the FCR responds positively to a stimulus … view at source ↗
Figure 2
Figure 2. (a) The decline on accuracy with increasing sym￾metric label swap noise on MiniImageNet; (b) The decline on accuracy with increasing symmetric label swap noise on TieredImageNet; (c) The decline on accuracy with increasing outlier noise on MiniImageNet; (d) The decline on accuracy with increasing outlier noise on TieredImageNet. is 60% and a higher proportion would obscure clean cate￾gories or reduce them to a minor… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

35 extracted references · 35 canonical work pages

  1. [25]

    Vinyals,O.,Blundell,C.,Lillicrap,T.,Kavukcuoglu,K.,Wierstra,D.,

    work page

  2. [18]

    Meta-learning for semi- supervisedfew-shotclassification,in:ProceedingsoftheInternational Conference on Learning Representations (ICLR)

    Ren,M.,Ravi,S.,Triantafillou,E.,Snell,J.,Swersky,K.,Tenenbaum, J.B., Larochelle, H., Zemel, R.S., 2018. Meta-learning for semi- supervisedfew-shotclassification,in:ProceedingsoftheInternational Conference on Learning Representations (ICLR)

    work page 2018

  3. [1]

    Me-momentum: Extracting hard confident examplesfromnoisilylabeleddata,in:ProceedingsoftheIEEE/CVF International Conference on Computer Vision (ICCV), pp

    Bai, Y., Liu, T., 2021. Me-momentum: Extracting hard confident examplesfromnoisilylabeleddata,in:ProceedingsoftheIEEE/CVF International Conference on Computer Vision (ICCV), pp. 9292– 9301

    work page 2021

  4. [2]

    An image is worth 16x16 words: Transformers for image recognition at scale, in: Proceedings of the International Conference on Learning Representations (ICLR)

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

    work page 2021

  5. [3]

    Context-awaremeta-learning,in:Proceedingsofthe International Conference on Learning Representations (ICLR)

    Fifty, C., Duan, D., Junkins, R.G., Amid, E., Leskovec, J., Re, C., Thrun,S.,2024. Context-awaremeta-learning,in:Proceedingsofthe International Conference on Learning Representations (ICLR)

    work page 2024

  6. [4]

    Co-teaching:robusttrainingofdeepneuralnet- workswithextremelynoisylabels,in:ProceedingsoftheConference onNeuralInformationProcessingSystems(NeurIPS),p.8536–8546

    Han, B., Yao, Q., Yu, X., Niu, G., Xu, M., Hu, W., Tsang, I.W., Sugiyama,M.,2018. Co-teaching:robusttrainingofdeepneuralnet- workswithextremelynoisylabels,in:ProceedingsoftheConference onNeuralInformationProcessingSystems(NeurIPS),p.8536–8546

    work page 2018

  7. [5]

    Safe-student for safe deep semi-supervised learning with unseen-class unlabeled data, in: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp

    He, R., Han, Z., Lu, X., Yin, Y., 2022a. Safe-student for safe deep semi-supervised learning with unseen-class unlabeled data, in: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 14565–14574

    work page 2022

  8. [6]

    He, Y., Liang, W., Zhao, D., Zhou, H.Y., Ge, W., Yu, Y., Zhang, W., 2022b. Attribute surrogates learning and spectral tokens pooling in transformers for few-shot learning, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 9109–9119

    work page

Show all 35 references
  1. [7]

    Rethinking generalization in few-shot classification, in: Proceedings of the Con- ference on Neural Information Processing Systems (NeurIPS), pp

    Hiller, M., Ma, R., Harandi, M., Drummond, T., 2024. Rethinking generalization in few-shot classification, in: Proceedings of the Con- ference on Neural Information Processing Systems (NeurIPS), pp. 3582–3595

    2024

  2. [8]

    General- purpose in-context learning by meta-learning transformers, in: Pro- ceedings of the Conference on Neural Information Processing Sys- tems (NeurIPS)

    Kirsch, L., Harrison, J., Sohl-Dickstein, J., Metz, L., 2022. General- purpose in-context learning by meta-learning transformers, in: Pro- ceedings of the Conference on Neural Information Processing Sys- tems (NeurIPS)

    2022

  3. [9]

    Dividemix:Learningwithnoisyla- belsassemi-supervisedlearning,in:ProceedingsoftheInternational Conference on Learning Representations (ICLR)

    Li,J.,Socher,R.,Hoi,S.C.,2020. Dividemix:Learningwithnoisyla- belsassemi-supervisedlearning,in:ProceedingsoftheInternational Conference on Learning Representations (ICLR)

    2020

  4. [10]

    Few- shot learning with noisy labels, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp

    Liang, K.J., Rangrej, S.B., Petrovic, V., Hassner, T., 2022. Few- shot learning with noisy labels, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 9079–9088

    2022

  5. [11]

    Early-learning regularization prevents memorization of noisy labels, in: Proceedings of the Conference on Neural Information Processing Systems (NeurIPS), pp

    Liu, S., Niles-Weed, J., Razavian, N., Fernandez-Granda, C., 2020. Early-learning regularization prevents memorization of noisy labels, in: Proceedings of the Conference on Neural Information Processing Systems (NeurIPS), pp. 20331–20342

    2020

  6. [12]

    Peerlossfunctions:learningfromnoisylabels without knowing noise rates, in: Proceedings of the International Conference on Machine Learning (ICML), pp

    Liu,Y.,Guo,H.,2020. Peerlossfunctions:learningfromnoisylabels without knowing noise rates, in: Proceedings of the International Conference on Machine Learning (ICML), pp. 6226–6236

    2020

  7. [13]

    Ma, X., Huang, H., Wang, Y., Romano, S., Erfani, S.M., Bailey, J.,

  8. [14]

    Rnnp: A robust few-shot learning approach, in: 2021 IEEE Winter Conference on Applications of Computer Vision (WACV), pp

    Mazumder, P., Singh, P., Namboodiri, V.P., 2021. Rnnp: A robust few-shot learning approach, in: 2021 IEEE Winter Conference on Applications of Computer Vision (WACV), pp. 2663–2672

    2021

  9. [15]

    A simple neural attentive meta-learner, in: Proceedings of the International Conference on Learning Representations (ICLR)

    Mishra, N., Rohaninejad, M., Chen, X., Abbeel, P., 2017. A simple neural attentive meta-learner, in: Proceedings of the International Conference on Learning Representations (ICLR)

    2017

  10. [16]

    Pervasivelabelerrors in test sets destabilize machine learning benchmarks, in: Proceed- ings of the Conference on Neural Information Processing Systems (NeurIPS)

    Northcutt,C.G.,Athalye,A.,Mueller,J.,2021. Pervasivelabelerrors in test sets destabilize machine learning benchmarks, in: Proceed- ings of the Conference on Neural Information Processing Systems (NeurIPS)

    2021

  11. [17]

    Coordinated drift of receptive fields in Hebbian/anti-Hebbian network models during noisy representation learning

    Qin, S., Farashahi, S., Lipshutz, D., Sengupta, A.M., Chklovskii, D.B., Pehlevan, C., 2023. Coordinated drift of receptive fields in Hebbian/anti-Hebbian network models during noisy representation learning. Nature Neuroscience 26, 339–349

    2023

  12. [19]

    Learning with symmetric label noise: the importance of being unhinged, in: Proceedings of the Conference on Neural Information Processing Systems (NeurIPS), pp

    Rooyen, B.v., Menon, A.K., Williamson, R.C., 2015. Learning with symmetric label noise: the importance of being unhinged, in: Proceedings of the Conference on Neural Information Processing Systems (NeurIPS), pp. 10–18

    2015

  13. [20]

    The balanced-pairwise-affinities feature transform, in: Proceedings of the International Conference on Machine Learning (ICML), pp

    Shalam, D., Korman, S., 2024. The balanced-pairwise-affinities feature transform, in: Proceedings of the International Conference on Machine Learning (ICML), pp. 44342–44357

    2024

  14. [21]

    Computational models link cellular mechanisms of neuromodulation to large-scale neural dynamics

    Shine, J.M., Müller, E.J., Munn, B., Cabral, J., Moran, R.J., Break- spear, M., 2021. Computational models link cellular mechanisms of neuromodulation to large-scale neural dynamics. Nature Neuro- science 24, 765–776

    2021

  15. [22]

    Prototypical networks for few-shot learning, in: Proceedings of the Conference on Neural Information Processing Systems (NeurIPS), pp

    Snell, J., Swersky, K., Zemel, R., 2017. Prototypical networks for few-shot learning, in: Proceedings of the Conference on Neural Information Processing Systems (NeurIPS), pp. 4080–4090

    2017

  16. [23]

    Learning to rectifyforrobustlearningwithnoisylabels

    Sun, H., Guo, C., Wei, Q., Han, Z., Yin, Y., 2022. Learning to rectifyforrobustlearningwithnoisylabels. PatternRecognition124, 108467

    2022

  17. [24]

    From imagenet to image classification: Contextualizing progress on benchmarks, in: Proceedings of the International Conference on Ma- chine Learning (ICML), pp

    Tsipras, D., Santurkar, S., Engstrom, L., Ilyas, A., Madry, A., 2020. From imagenet to image classification: Contextualizing progress on benchmarks, in: Proceedings of the International Conference on Ma- chine Learning (ICML), pp. 9625 – 9635

    2020

  18. [26]

    Wang,X.,Hua,Y.,Kodirov,E.,Clifton,D.A.,Robertson,N.M.,2021. Proselflc: Progressive self label correction for training robust deep neural networks, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 752–761

    2021

  19. [27]

    Wei, H., Feng, L., Chen, X., An, B., 2020. Combating noisy labels by agreement: A joint training method with co-regularization, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 13723–13732

    2020

  20. [28]

    Fine-grained classification with noisy labels, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp

    Wei, Q., Feng, L., Sun, H., Wang, R., Guo, C., Yin, Y., 2023. Fine-grained classification with noisy labels, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 11651–11660

    2023

  21. [29]

    Learningtopurify noisy labels via meta soft label corrector

    Wu,Y.,Shu,J.,Xie,Q.,Zhao,Q.,Meng,D.,2021. Learningtopurify noisy labels via meta soft label corrector. Proceedings of the AAAI Conference on Artificial Intelligence 35, 10388–10396

    2021

  22. [30]

    Object detection as a positive- unlabeled problem, in: Proceedings of the British Machine Vision Conference (BMVC)

    Yang, Y., Liang, K., Carin, L., 2020. Object detection as a positive- unlabeled problem, in: Proceedings of the British Machine Vision Conference (BMVC)

    2020

  23. [31]

    Understandingdeeplearning(still)requiresrethinkinggeneralization

    Zhang, C., Bengio, S., Hardt, M., Recht, B., Vinyals, O., 2021a. Understandingdeeplearning(still)requiresrethinkinggeneralization. Communications of the ACM 64, 107–115

  24. [32]

    Shallow bayesian meta learning for real-world few-shot recognition

    Zhang, X., Meng, D., Gouk, H.G.R., Hospedales, T.M., 2021b. Shallow bayesian meta learning for real-world few-shot recognition. ProceedingsoftheIEEE/CVFInternationalConferenceonComputer Vision (ICCV) , 631–640

  25. [33]

    Meta label correction for noisy label learning

    Zheng, G., Hassan, A., Dumais, S., 2021. Meta label correction for noisy label learning. Proceedings of the AAAI Conference on Artificial Intelligence 35, 11053–11061. Nan Xiang et al.:Preprint submitted to ElsevierPage 7 of 7

    2021

  26. [2016]

    3637–3645

    Matchingnetworksforoneshotlearning,in:Proceedingsofthe ConferenceonNeuralInformationProcessingSystems(NeurIPS),pp. 3637–3645

  27. [2020]

    6543–6553

    Normalized loss functions for deep learning with noisy labels, in:ProceedingsoftheInternationalConferenceonMachineLearning (ICML), pp. 6543–6553

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.