Pith. sign in

REVIEW 4 major objections 5 minor 231 references

Leveraging Dissimilarity Invariance as a Robust Anchor for Learning with Noisy Labels

T0 review · 4 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

Pith's one-line read Label noise distorts similarity between related images but leaves dissimilarity between unrelated images stable, and that stability can be used as a robust anchor for training under noisy supervision.

desk verdict The dissimilarity-anchor idea is a genuinely useful empirical framing and the plug-in losses look effective, but the theory is not credible as written and the negative-pair set built from noisy labels needs more scrutiny. read the letter →

arxiv 2607.17857 v1 pith:EW6CE667 submitted 2026-07-20 cs.CV

classification cs.CV
keywords noisylabelsdissimilarityinvariancenegativepairsorthogonalitypenaltysimilaritycalibrationpLCAdistancerepresentationlearninglabelnoiserobustness
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

This paper tries to establish that under label noise, the semantic gap between unrelated samples—their dissimilarity—remains stable, even as similarity between related samples is badly distorted. If true, that stable signal gives training a reliable anchor that existing noisy-label methods lack, because they build on fragile similarities. The authors propose NegScale, a plug-in framework with two losses: one pushes negative (unrelated) pairs toward orthogonal directions, and one uses those dissimilarity anchors to cap spurious similarity among positive pairs. They report consistent gains over prior methods on synthetic and real-world noisy benchmarks.

What carries the argument

The carrying mechanism is the negative-pair set P_neg: sample pairs whose pLCA distance—a taxonomic tree-distance between their (noisy) class labels—exceeds a threshold η_max, taken to be semantically unrelated. Dissimilarity Invariance says these pairs' similarities are stable under noise, so they serve as a trustworthy anchor. SNOP enforces orthogonality among those pairs' difference vectors, and DCSA uses them to compute an adaptive upper bound on positive-pair similarity. The pLCA distance with hand-chosen thresholds is what defines the anchor; everything else is built on it.

What would settle it

Measure, under each noise type, the precision of P_neg: the fraction of pairs in the negative set whose ground-truth labels differ, as noise rate rises. If that precision drops sharply, or if a direct first-order measurement shows negative-pair similarity changes as much as positive-pair similarity, the Dissimilarity Invariance anchor is not doing the claimed work.

Watch

Extended reading notes

Core claim

The paper's central discovery is the phenomenon of Dissimilarity Invariance: the cosine similarity between samples from semantically unrelated classes stays nearly constant as label noise increases, while similarity between related or same-class samples shifts strongly (Table 1, Figs. 2–3). The authors argue label noise corrupts the positive structure but leaves the negative structure intact, making dissimilarity a trustworthy anchor. On that basis they design two complementary mechanisms: SNOP, which penalizes deviation from orthogonality among negative pairs (both globally—difference vectors forming an identity Gram matrix—and locally with confidence weighting), and DCSA, which for each po

Load-bearing premise

The whole scheme assumes that the set of 'unrelated' pairs, built from the same noisy labels with fixed pLCA thresholds, is actually dominated by truly unrelated classes; if noise is strong enough or structured enough to put same-class samples in that set, the stable anchor is no longer stable.

Editorial extensions

If this is right

  • If dissimilarity is invariant, noisy-label training can be anchored on negative structure rather than positive structure, which most prior methods rely on.
  • The plug-in design means existing noisy-label methods (e.g., RoLR, DivideMix, RankMatch, ANNE) can gain by adding SNOP and DCSA, as shown in Table 6.
  • Under heavy symmetric noise (80%), the gap over strong baselines grows, suggesting the anchor matters most when the positive signal is most degraded.
  • The theoretical first-order analysis predicts that per-gradient-step similarity change for negative pairs is smaller than for positive pairs; if that holds, the invariance is a dynamical property of CE-SGD, not just a dataset artifact.

Reading between the lines

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

  • The invariance claim is tested on image benchmarks with class taxonomies; a natural extension is to domains where class structure is flatter or hierarchical in different ways, where pLCA-based negative selection may need re-derivation.
  • Because the anchor is defined from noisy labels, the method's benefit may depend on the noise being roughly class- or instance-independent; under adversarial or extreme asymmetric noise that flips same-class samples into the negative set, the anchor itself becomes contaminated—a regime not measured in the paper.
  • The orthogonality constraint on negative pairs could interact with representation dimensionality; in low-dimensional embeddings, enforcing exact orthogonality on many pairs may be infeasible, suggesting an adaptive target rank as a testable extension.
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 / 5 minor

Summary. The paper observes that, under label noise, pairwise cosine similarities between semantically unrelated samples ("dissimilarity") remain more stable than similarities between semantically related samples. It names this phenomenon Dissimilarity Invariance and proposes NegScale, a plug-and-play regularizer with two terms: SNOP, which enforces orthogonality between features of negative pairs, and DCSA, which suppresses positive-pair similarity when one member is close to a negative anchor. The authors provide a theoretical analysis intended to prove Dissimilarity Invariance under symmetric noise and to bound the generalization error of NegScale (Lemma 1, Theorem 1). Empirically, NegScale combined with RoLR is evaluated on CIFAR-10/100 with synthetic symmetric, asymmetric, and instance-dependent noise, on CIFAR-10N/100N, Animal-10N, and WebVision, and is reported to outperform prior state-of-the-art methods. Ablations show that both SNOP and DCSA contribute to the gains.

Significance. If Dissimilarity Invariance holds as stated, it offers a conceptually simple and potentially useful principle: instead of trying to repair fragile positive-pair similarities under label noise, one can anchor learning on stable negative-pair dissimilarities. The proposed NegScale is architecture-agnostic and the reported empirical gains are consistent across many settings, including real-world noisy datasets, which is a useful practical contribution. The paper also includes a welcome sensitivity analysis and ablations. However, the theoretical justification is currently not rigorous: the invariance proof contains an unjustified term drop and an apparent logical reversal, and Lemma 1/Theorem 1 rely on an unprovided appendix and a suspicious denominator. In addition, the negative-pair set P_neg is constructed from noisy labels, so the central anchor may itself be corrupted under asymmetric or instance-dependent noise. These issues do not necessarily invalidate the empirical claims, but they need to be addressed before the paper can be accepted.

major comments (4)
  1. [Theoretical Analysis, Eqs. (10)-(14)] The proof of Dissimilarity Invariance is not valid as written. In Eq. (13), the term p_j^{(\tilde y_i)} is dropped with the justification that it is small for yi != yj, but under label noise \tilde y_i can equal y_j, and even when it does not, this term is precisely what encodes class overlap. The subsequent Eq. (14) writes |Δs_ij| - |Δs_i'j'| = |2γ(Σ p_i p_j - Σ p_i' p_j')|, which is algebraically incorrect after dropping the negative term, since the absolute value does not distribute. More importantly, the text states that semantically similar pairs have smaller Kendall tau distance than negative pairs, but then says "Kendall tau distance of (i,j) is greater than that of (i',j')", reversing the ordering; the rearrangement-inequality step is therefore not justified. The proof assumes the ordering it needs to establish. Please correct the derivation or state the required assumption expli
  2. [Lemma 1 and Theorem 1, Eqs. (15) and (18)] The proofs are deferred to an appendix that is not present in the manuscript. More concerning, the bound in Eq. (15) has denominator λ·δ_SNOP + μ·δ_DCSA. If δ_SNOP and δ_DCSA are the minimized residuals of the regularizers, then a smaller residual (i.e., better regularization) makes the denominator smaller and the bound worse, which is counterintuitive. If they are intended as something else, that is not defined. Also, the assumptions that L_SNOP and L_DCSA can be minimized to at most δ_SNOP and δ_DCSA, and that the CE gradient norm is bounded by G, are stated without any construction or discussion. As written, the lemma and theorem do not provide a meaningful quantitative guarantee. Please provide complete proofs, clarify the role of the δs, and ensure the bound behaves sensibly as regularization improves.
  3. [Methodology, Eq. (1) and P_neg construction] The negative-pair set P_neg is defined using pLCA distances computed from the noisy labels: (i,j) ∈ P_neg iff D_pLCA(\tilde y_i, \tilde y_j) > η_max. Under asymmetric or instance-dependent noise, same-class samples can receive noisy labels that are far apart in the taxonomy and therefore enter P_neg; once this happens, SNOP actively pushes same-class representations apart and DCSA uses corrupted anchors. The paper claims robustness on Pair and Ins noise (Table 2) but the theoretical analysis covers only symmetric noise, and no experiment reports the purity of P_neg under each noise type. Since the entire method rests on P_neg being a reliable dissimilarity anchor, this is a load-bearing gap. Please report P_neg precision (fraction of pairs that are truly unrelated) under each noise setting, and discuss how the method degrades when P_neg is corrupted.
  4. [Experimental protocol, Tables 2-5] The main comparisons borrow baseline numbers from prior papers under mixed training protocols, and no error bars, standard deviations, or number of seeds are reported. The claim of "consistently outperforms state-of-the-art" would be stronger with a unified protocol and variance estimates. At minimum, please report the mean±std over at least three runs for the proposed method and re-run the key baselines under the same evaluation setup. This is important because some reported gains, e.g., CIFAR-100 Pair 77.8 vs. 76.1, may be within run-to-run variability.
minor comments (5)
  1. [Throughout] Typos and inconsistent notation: "Releated Work", "nutshull", "benigning" (in Eq. 14 discussion), and "DSCA" in Table 7 (should be DCSA). Please proofread.
  2. [Figures 2-3] The captions mention "Small/Large Clean" and "Small/Large pLCA", but the legend is not described in text; clarify what these terms denote and how the sets are constructed.
  3. [Eq. (5)] The confidence weight w_ij uses per-sample confidence c_i, but the definition of c_i is informal ("softmax probability of its predicted class"). Specify whether this is computed with the noisy label or the model's predicted class, and how it is aggregated over epochs.
  4. [Section: Key Observations] Table 1 reports "Mean" and "Δ With Clean" but the rows for real-world noise are separated from synthetic noise; the formatting makes it hard to see which rows contribute to the mean. Clarify.
  5. [Appendix] The paper refers to an appendix for the pLCA hierarchy, noise injection details, and proofs, but the appendix is not included in the submitted manuscript. Please include it.

Circularity Check

2 steps flagged · score 6.0 of 10

Invariance proof assumes its conclusion; effectiveness bound is tied to its own losses.

  1. self definitional [Theoretical Analysis, 'Why Negative-Pair Similarity Remains Invariant', Eqs. (10)-(14)]
    "From the nature of semantic similarity, we know that semantically related pairs (i, j)∈P_sim tend to have a higher overlap in their logits ... while semantically unrelated pairs (i′, j′)∈P_neg exhibit minimal overlap. ... From the benigning, we know that the Kendall tau distance of (i,j) is greater than that of (i′,j′). According to the rearrangement inequality (Cvetkovski 2012), this implies that Eq. (14)>0."

    The theorem aims to prove that dissimilarity (negative-pair similarity) is more invariant than related-pair similarity under label noise. But Eq. (14) reduces |Δs_ij| − |Δs_i′j′| exactly to the difference Σ p_i^(k)p_j^(k) − Σ p_i′^(k)p_j′^(k). The proof supplies no independent derivation of the sign of this difference; it merely asserts, as 'the nature of semantic similarity,' that related pairs have higher logit overlap (smaller Kendall tau distance). That ordering is equivalent to the desired inequality, so the proof restates its conclusion. The dropped term p_j^(ỹ_i) also presupposes the discriminative behavior under investigation.

  2. other [Theoretical Analysis, 'Why NegScale is Effective', Lemma 1 (Eq. (15)) and Theorem 1]
    "Assume each of L_SNOP and L_DCSA is minimized to at most δ_SNOP, δ_DCSA respectively, and that the gradient norm of cross-entropy loss on noisy labels is bounded by ∥∇L_CE∥ ≤ G. Then the feature perturbation due to noisy labels is upper bounded as: ∥f̃_i − f_i∥ ≤ G·τ/(λ·δ_SNOP + μ·δ_DCSA)."

    The claimed robustness bound is tied to its own objective by construction: the denominator contains the very quantities δ_SNOP and δ_DCSA that are assumed as upper bounds on the SNOP and DCSA losses. Making those assumed bounds smaller makes the denominator smaller and hence the bound larger, so the theorem cannot establish that minimizing NegScale suppresses perturbation; it only restates the assumption that these regularized losses control the perturbation. No proof is supplied (deferred to an absent appendix), so the 'effectiveness' conclusion reduces to an unverified assumption about the same losses being minimized.

full rationale

The paper's empirical section is largely self-contained: NegScale is compared against external baselines on CIFAR, CIFAR-N, Animal-10N and WebVision, and the plug-in/ablation results are independently meaningful. I found no load-bearing self-citation: Fan & Li 2025 (CCL) is cited as a baseline/related method, not as the justification for Dissimilarity Invariance or NegScale. However, the theoretical justification contains two self-referential steps. The proof of Dissimilarity Invariance (Eqs. 10-14) assumes the exact logit-overlap ordering it needs to prove; Eq. (14) reduces the claimed inequality to the dot-product ordering supplied as 'the nature of semantic similarity.' Lemma 1/Theorem 1 then state that minimizing SNOP/DCSA bounds feature perturbation, but the bound's denominator is the assumed upper bounds δ_SNOP and δ_DCSA of those very losses, and the proof is deferred to an absent appendix. These steps make the theoretical derivation partially circular, while the empirical benchmarks still carry independent weight. The P_neg/noisy-label-purity concern is a correctness risk, not a circularity per se.

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

No new physical entities are introduced. The framework's free parameters are the loss weights and pLCA thresholds, which are chosen by hand and tuned via sensitivity analysis. The theoretical analysis depends on several ad hoc assumptions about logit overlap, discriminative ability, and taxonomy correctness that are not independently established.

free parameters (4)
  • lambda (SNOP weight) = 1 (default)
    Chosen by hand in Implementation Details; sensitivity analysis in Fig. 5(a) shows (1,1) is best, so it is tuned on validation.
  • mu (DCSA weight) = 1 (default)
    Chosen by hand in Implementation Details; sensitivity analysis in Fig. 5(a) shows (1,1) is best, so it is tuned on validation.
  • eta_min (pLCA relatedness lower threshold) = 3
    Hand-selected threshold for deciding which pairs are semantically related; sensitivity analysis in Fig. 5(b).
  • eta_max (pLCA relatedness upper threshold) = 5
    Hand-selected threshold for deciding which pairs are semantically unrelated; sensitivity analysis in Fig. 5(b).
assumptions (4)
  • ad hoc to paper Semantically related pairs have higher overlap in classifier logits (z_i^k ≈ z_j^k for most k), while unrelated pairs have minimal overlap.
    This is the premise of the Dissimilarity Invariance proof (Theoretical Analysis); it is essentially the claim being proven, not an independently verified fact.
  • ad hoc to paper The dropped term p_j^{y~i} is negligible when y_i != y_j for a model with 'reasonable discriminative ability'.
    Invoked to pass from Eq. (13) to the approximation; under high symmetric noise discriminative ability is compromised, so the approximation may fail.
  • domain assumption The WordNet/pLCA class taxonomy used to define P_neg accurately reflects semantic relatedness for all datasets evaluated.
    P_neg drives both losses; the taxonomy is deferred to the Appendix and is not justified for CIFAR/WebVision classes.
  • ad hoc to paper Lemma 1 assumes L_SNOP and L_DCSA can be minimized to at most delta_SNOP and delta_DCSA, and that the CE gradient norm is bounded by G; no construction of such bounds is provided.
    The lemma and Theorem 1 rely on unproved constants and a proof deferred to an absent appendix.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Leveraging Dissimilarity Invariance as a Robust Anchor for Learning with Noisy Labels." pith.science (2026). https://pith.science/paper/EW6CE667

@misc{pith2026260717857,
  author       = {Pith},
  title        = {Pith review of: Leveraging Dissimilarity Invariance as a Robust Anchor for Learning with Noisy Labels},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EW6CE667}},
  note         = {Machine review of arXiv:2607.17857}
}
read the original abstract

Deep learning models excel in visual recognition but suffer severe performance drops when training labels are corrupted by noise. Under label noise prior work cannot learn accurate similarities and thus misguide the learning process. In this paper, we uncover a complementary and novel phenomenon, Dissimilarity Invariance, whereby semantic dissimilarity between unrelated samples remains stable despite label noise. Leveraging this insight, we propose NegScale, a plug-and-play framework that shifts focus from fragile similarity to robust dissimilarity. NegScale integrates: (1) Structured Negative Orthogonality Penalty (SNOP), enforcing subspace orthogonality for unrelated samples; and (2) Dissimilarity-Calibrated Similarity Adjustment (DCSA), suppressing spurious similarity using dissimilarity anchors. We also give theoretical analysis that proves Dissimilarity Invariance and the effectiveness of NegScale. Empirical results demonstrate that NegScale consistently outperforms state-of-the-art baselines, establishing new benchmarks on CIFAR with synthetic noise and real-world datasets.

Figures

Figures reproduced from arXiv: 2607.17857 by the authors.

Figure 1
Figure 1. Illustration of Dissimilarity Invariance. Semantic [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Key observations of dissimilarity invariance under synthetic noises with CIFAR-10. Small / Large pLCA refers to [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Key observations of dissimilarity invariance under various noise types of real world noisy dataset CIFAR-10N. Small [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Results on similarity variation after training with [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Ablation study on hyperparameters with CIFAR [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

231 extracted references · 15 canonical work pages

  1. [1]

    FirstName LastName , title =

  2. [2]

    FirstName Alpher , title =

  3. [3]

    Journal of Foo , volume = 13, number = 1, pages =

    FirstName Alpher and FirstName Fotheringham-Smythe , title =. Journal of Foo , volume = 13, number = 1, pages =

  4. [4]

    Journal of Foo , volume = 14, number = 1, pages =

    FirstName Alpher and FirstName Fotheringham-Smythe and FirstName Gamow , title =. Journal of Foo , volume = 14, number = 1, pages =

  5. [5]

    FirstName Alpher and FirstName Gamow , title =

  6. [6]

    Abril and Robert Plant

    Patricia S. Abril and Robert Plant. The patent holder's dilemma: Buy, sell, or troll?. Communications of the ACM. 2007. doi:10.1145/1188913.1188915

  7. [7]

    Deciding equivalances among conjunctive aggregate queries

    Sarah Cohen and Werner Nutt and Yehoshua Sagic. Deciding equivalances among conjunctive aggregate queries. doi:10.1145/1219092.1219093

  8. [8]

    Special issue: Digital Libraries. 1996

Show all 231 references
  1. [9]

    Understanding Policy-Based Networking

    David Kosiur. Understanding Policy-Based Networking. 2001

  2. [12]

    The title of book two. 2008. doi:10.1007/3-540-09237-4

  3. [13]

    Asad Z. Spector. Achieving application requirements. Distributed Systems. 1990. doi:10.1145/90417.90738

  4. [14]

    Douglass and David Harel and Mark B

    Bruce P. Douglass and David Harel and Mark B. Trakhtenbrot. Statecarts in use: structured analysis and object-orientation. Lectures on Embedded Systems. 1998. doi:10.1007/3-540-65193-4_29

  5. [15]

    Donald E. Knuth. The Art of Computer Programming, Vol. 1: Fundamental Algorithms (3rd. ed.). 1997

  6. [16]

    Donald E. Knuth. The Art of Computer Programming. 1998

  7. [17]

    Structured Variational Inference Procedures and their Realizations (as incol)

    Dan Geiger and Christopher Meek. Structured Variational Inference Procedures and their Realizations (as incol). Proceedings of Tenth International Workshop on Artificial Intelligence and Statistics, The Barbados

  8. [18]

    Stan W. Smith. An experiment in bibliographic mark-up: Parsing metadata for XML export. Proceedings of the 3rd. annual workshop on Librarians and Computers. 2010. doi:99.9999/woot07-S422

  9. [19]

    Catch me, if you can: Evading network signatures with web-based polymorphic worms

    Matthew Van Gundy and Davide Balzarotti and Giovanni Vigna. Catch me, if you can: Evading network signatures with web-based polymorphic worms. Proceedings of the first USENIX workshop on Offensive Technologies

  10. [20]

    Predicate Path expressions

    Sten Andler. Predicate Path expressions. Proceedings of the 6th. ACM SIGACT-SIGPLAN symposium on Principles of Programming Languages. 1979. doi:10.1145/567752.567774

  11. [21]

    LOGICS of Programs: AXIOMATICS and DESCRIPTIVE POWER

    David Harel. LOGICS of Programs: AXIOMATICS and DESCRIPTIVE POWER. 1978

  12. [22]

    Anisi , title =

    David A. Anisi , title =

  13. [23]

    Clarkson

    Kenneth L. Clarkson. Algorithms for Closest-Point Problems (Computational Geometry). 1985

  14. [24]

    Introduction to Bayesian Statistics

    Harry Thornburg. Introduction to Bayesian Statistics. 2001

  15. [25]

    CLIFFORD: a Maple 11 Package for Clifford Algebra Computations, version 11

    Rafal Ablamowicz and Bertfried Fauser. CLIFFORD: a Maple 11 Package for Clifford Algebra Computations, version 11. 2007

  16. [26]

    Stats and Analysis

    Poker-Edge.Com. Stats and Analysis. 2006

  17. [27]

    A more perfect union

    Barack Obama. A more perfect union. 2008

  18. [28]

    The fountain of youth

    Joseph Scientist. The fountain of youth. 2009

  19. [29]

    Solder man

    Dave Novak. Solder man. ACM SIGGRAPH 2003 Video Review on Animation theater Program: Part I - Vol. 145 (July 27--27, 2003). doi:99.9999/woot07-S422

  20. [30]

    Interview with Bill Kinder: January 13, 2005

    Newton Lee. Interview with Bill Kinder: January 13, 2005. Comput. Entertain. 2005. doi:10.1145/1057270.1057278

  21. [31]

    The Enabling of Digital Libraries

    Bernard Rous. The Enabling of Digital Libraries. Digital Libraries. 2008

  22. [33]

    (new) Finding minimum congestion spanning trees , journal =

    Werneck, Renato and Setubal, Jo\. (new) Finding minimum congestion spanning trees , journal =. doi:10.1145/351827.384253 , acmid = 384253, publisher =

  23. [35]

    and Mei, Alessandro , title =

    Conti, Mauro and Di Pietro, Roberto and Mancini, Luigi V. and Mei, Alessandro , title =. Inf. Fusion , volume =. 2009 , issn =. doi:10.1016/j.inffus.2009.01.002 , acmid =

  24. [36]

    and Hutchful, David K

    Li, Cheng-Lun and Buyuktur, Ayse G. and Hutchful, David K. and Sant, Natasha B. and Nainwal, Satyendra K. , title =. CHI '08 extended abstracts on Human factors in computing systems , year =. doi:10.1145/1358628.1358946 , acmid =

  25. [37]

    , title =

    Hollis, Billy S. , title =. 1999 , isbn =

  26. [38]

    Goossens, Michel and Rahtz, S. P. and Moore, Ross and Sutor, Robert S. , title =. 1999 , isbn =

  27. [39]

    and Rosenberg, Arnold L

    Buss, Jonathan F. and Rosenberg, Arnold L. and Knott, Judson D. , title =. 1987 , source =

  28. [40]

    CHI '08: CHI '08 extended abstracts on Human factors in computing systems , year =

    , note =. CHI '08: CHI '08 extended abstracts on Human factors in computing systems , year =

  29. [41]

    Algorithms for Closest-Point Problems (Computational Geometry) , year =

    Clarkson, Kenneth Lee , advisor =. Algorithms for Closest-Point Problems (Computational Geometry) , year =

  30. [42]

    SIGCOMM Comput. Commun. Rev. , year =

  31. [43]

    2004 , isbn =

    IEEE TCSC Executive Committee , booktitle =. 2004 , isbn =. doi:http://dx.doi.org/10.1109/ICWS.2004.64 , acmid =

  32. [44]

    Distributed systems (2nd Ed.) , year =

  33. [45]

    , title =

    Petrie, Charles J. , title =. 1986 , source =

  34. [46]

    Donald E. Knuth. Seminumerical Algorithms. 1981

  35. [47]

    E-commerce and cultural values , year =

    Kong, Wei-Chang , Title =. E-commerce and cultural values , year =

  36. [48]

    E-commerce and cultural values , year =

    Kong, Wei-Chang , type =. E-commerce and cultural values , year =

  37. [49]

    Chapter 9 , booktitle =

    Kong, Wei-Chang , editor =. Chapter 9 , booktitle =. 2002 , address =

  38. [50]

    E-commerce and cultural values , editor =

    Kong, Wei-Chang , title =. E-commerce and cultural values , editor =. 2003 , isbn =

  39. [51]

    E-commerce and cultural values - (InBook-num-in-chap) , chapter =

    Kong, Wei-Chang , editor =. E-commerce and cultural values - (InBook-num-in-chap) , chapter =. 2004 , address =

  40. [52]

    E-commerce and cultural values (Inbook-text-in-chap) , chapter =

    Kong, Wei-Chang , editor =. E-commerce and cultural values (Inbook-text-in-chap) , chapter =. 2005 , address =

  41. [53]

    E-commerce and cultural values (Inbook-num chap) , chapter =

    Kong, Wei-Chang , editor =. E-commerce and cultural values (Inbook-num chap) , chapter =. 2006 , address =

  42. [54]

    Microelectron

    Mehdi Saeedi and Morteza Saheb Zamani and Mehdi Sedighi , title =. Microelectron. J. , volume =. 2010 , pages =

  43. [55]

    Mehdi Saeedi and Morteza Saheb Zamani and Mehdi Sedighi and Zahra Sasanian , title =. J. Emerg. Technol. Comput. Syst. , volume =

  44. [56]

    Kirschmer, Markus and Voight, John , title =. SIAM J. Comput. , issue_date =. 2010 , issn =. doi:https://doi.org/10.1137/080734467 , acmid =

  45. [57]

    Hoare, C. A. R. , title =. Structured programming (incoll) , editor =. 1972 , isbn =

  46. [58]

    History of programming languages I (incoll) , editor =

    Lee, Jan , title =. History of programming languages I (incoll) , editor =. 1981 , isbn =. doi:http://doi.acm.org/10.1145/800025.1198348 , acmid =

  47. [59]

    , title =

    Dijkstra, E. , title =. Classics in software engineering (incoll) , year =

  48. [60]

    , title =

    Wenzel, Elizabeth M. , title =. Multimedia interface design (incoll) , year =. doi:10.1145/146022.146089 , acmid =

  49. [61]

    , title =

    Mumford, E. , title =. Critical issues in information systems research (incoll) , year =

  50. [62]

    and Golden, Donald G

    McCracken, Daniel D. and Golden, Donald G. , title =. 1990 , isbn =

  51. [63]

    The analysis of linear partial differential operators

    H. The analysis of linear partial differential operators. 1985 , PAGES =

  52. [64]

    IEEE", address =

    A. Adya and P. Bahl and J. Padhye and A.Wolman and L. Zhou , title =. Proceedings of the IEEE 1st International Conference on Broadnets Networks (BroadNets'04) , publisher = "IEEE", address = "Los Alamitos, CA", year =

  53. [65]

    I. F. Akyildiz and W. Su and Y. Sankarasubramaniam and E. Cayirci , title =. Comm. ACM , volume = 38, number = "4", year =

  54. [66]

    I. F. Akyildiz and T. Melodia and K. R. Chowdhury , title =. Computer Netw. , volume = 51, number = "4", year =

  55. [67]

    ACM", address =

    P. Bahl and R. Chancre and J. Dungeon , title =. Proceeding of the 10th International Conference on Mobile Computing and Networking (MobiCom'04) , publisher = "ACM", address = "New York, NY", year =

  56. [68]

    8 (Special Issue on Sensor Networks)

    D. Culler and D. Estrin and M. Srivastava , title =. IEEE Comput. , volume = 37, number = "8 (Special Issue on Sensor Networks)", publisher = "IEEE", address = "Los Alamitos, CA", year =

  57. [69]

    Natarajan and M

    A. Natarajan and M. Motani and B. de Silva and K. Yap and K. C. Chua , title =. Network Architectures , editor =. 960935712

  58. [70]

    Tzamaloukas and J

    A. Tzamaloukas and J. J. Garcia-Luna-Aceves , title =

  59. [71]

    Zhou and J

    G. Zhou and J. Lu and C.-Y. Wan and M. D. Yarvis and J. A. Stankovic , title =

  60. [72]

    Mapping Powerlists onto Hypercubes

    Jacob Kornerup. Mapping Powerlists onto Hypercubes. 1994

  61. [73]

    Automatic Parallelization for Distributed-Memory Multiprocessing Systems

    Michael Gerndt. Automatic Parallelization for Distributed-Memory Multiprocessing Systems

  62. [74]

    J. E. Archer, Jr. and R. Conway and F. B. Schneider. User recovery and reversal in interactive systems. ACM Trans. Program. Lang. Syst

  63. [75]

    D. D. Dunlop and V. R. Basili. Generalizing specifications for uniformly implemented loops. ACM Trans. Program. Lang. Syst

  64. [76]

    Heering and P

    J. Heering and P. Klint. Towards monolingual programming environments. ACM Trans. Program. Lang. Syst

  65. [77]

    Donald E. Knuth. The book

  66. [78]

    Korach and D

    E. Korach and D. Rotem and N. Santoro. Distributed algorithms for finding centers and medians in networks. ACM Trans. Program. Lang. Syst

  67. [79]

    : A Document Preparation System

    Leslie Lamport. : A Document Preparation System

  68. [80]

    F. Nielson. Program transformations in a denotational setting. ACM Trans. Program. Lang. Syst

  69. [81]

    Brian K. Reid. A high-level approach to computer document formatting. Proceedings of the 7th Annual Symposium on Principles of Programming Languages

  70. [82]

    and Abdelzaher, Tarek F

    Zhou, Gang and Wu, Yafeng and Yan, Ting and He, Tian and Huang, Chengdu and Stankovic, John A. and Abdelzaher, Tarek F. , title =. ACM Trans. Embed. Comput. Syst. , issue_date =. doi:10.1145/1721695.1721705 , acmid = 1721705, publisher =

  71. [83]

    Institutional members of the Users Group

  72. [84]

    Boris Veytsman , title =

  73. [85]

    Robin Schneider , title =

  74. [86]

    and Peterson, Larry L

    Bowman, Mic and Debray, Saumya K. and Peterson, Larry L. , title =. ACM Trans. Program. Lang. Syst. , volume =. 1993 , doi =

  75. [87]

    TUGboat , volume =

    Braams, Johannes , title =. TUGboat , volume =

  76. [88]

    Post Congress Tristesse

    Malcolm Clark. Post Congress Tristesse. TeX90 Conference Proceedings

  77. [89]

    ACM Trans

    Herlihy, Maurice , title =. ACM Trans. Program. Lang. Syst. , volume =. 1993 , doi =

  78. [90]

    Salas and Einar Hille

    S.L. Salas and Einar Hille. Calculus: One and Several Variable. 1978

  79. [91]

    Publication quality tables in

    Simon Fear , month =. Publication quality tables in

  80. [92]

    Using the amsthm Package , organization =

  81. [93]

    2019 , url =

    R: A Language and Environment for Statistical Computing , author =. 2019 , url =

  82. [94]

    Sam Anzaroot and Andrew McCallum , title =

  83. [95]

    Brad and Haunschild, Robin , title =

    Bornmann, Lutz and Wray, K. Brad and Haunschild, Robin , title =

  84. [96]

    2014 , archivePrefix =

    Sam Anzaroot and Alexandre Passos and David Belanger and Andrew McCallum , title =. 2014 , archivePrefix =

  85. [97]

    Proceedings of the 20th International Colloquium on Automata, Languages and Programming , series =

    Maintaining Discrete Probability Distributions Optimally , author =. Proceedings of the 20th International Colloquium on Automata, Languages and Programming , series =. 1993 , publisher =

  86. [98]

    FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence , booktitle =

    Kihyuk Sohn and David Berthelot and Nicholas Carlini and Zizhao Zhang and Han Zhang and Colin Raffel and Ekin Dogus Cubuk and Alexey Kurakin and Chun. FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence , booktitle =. 2020 , url =

  87. [99]

    Journal of Machine Learning Research , year=

    Visualizing Data using t-SNE , author=. Journal of Machine Learning Research , year=

  88. [100]

    Enhanced Data-Recalibration: Utilizing Validation Data to Mitigate Instance-Dependent Noise in Classification

    Germi, Saeed Bakhshi and Rahtu, Esa. Enhanced Data-Recalibration: Utilizing Validation Data to Mitigate Instance-Dependent Noise in Classification. Image Analysis and Processing -- ICIAP 2022. 2022

  89. [101]

    Structure and Interpretation of Computer Programs

    Harold Abelson and Gerald Jay Sussman and Julie Sussman. Structure and Interpretation of Computer Programs. 1985

  90. [102]

    Robust Learning Against Label Noise Based on Activation Trend Tracking , year=

    Wang, Yilin and Zhang, Yulong and Jiang, Zhiqiang and Zheng, Li and Chen, Jinshui and Lu, Jiangang , journal=. Robust Learning Against Label Noise Based on Activation Trend Tracking , year=

  91. [103]

    Li , editor =

    Cheng Tan and Jun Xia and Lirong Wu and Stan Z. Li , editor =. Co-learning: Learning from Noisy Labels with Self-supervision , booktitle =. 2021 , url =. doi:10.1145/3474085.3475622 , timestamp =

  92. [104]

    Patel, Deep and Sastry, P. S. , year =. Adaptive. arXiv:2106.15292 [cs] , eprint =

  93. [105]

    Biometrics , volume=

    Residuals and Influence Regression , author=. Biometrics , volume=

  94. [106]

    Proceedings of the 34th International Conference on Machine Learning , pages =

    Understanding Black-box Predictions via Influence Functions , author =. Proceedings of the 34th International Conference on Machine Learning , pages =. 2017 , editor =

  95. [107]

    Erfani and James Bailey , title =

    Xingjun Ma and Hanxun Huang and Yisen Wang and Simone Romano and Sarah M. Erfani and James Bailey , title =. Proceedings of the 37th International Conference on Machine Learning,. 2020 , url =

  96. [108]

    and Tsvetkov, Yulia

    Han, Xiaochuang and Wallace, Byron C. and Tsvetkov, Yulia. Explaining Black Box Predictions and Unveiling Data Artifacts through Influence Functions. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. 2020. doi:10.18653/v1/2020.acl-main.492

  97. [109]

    Estimating Training Data Influence by Tracing Gradient Descent , url =

    Pruthi, Garima and Liu, Frederick and Kale, Satyen and Sundararajan, Mukund , booktitle =. Estimating Training Data Influence by Tracing Gradient Descent , url =

  98. [110]

    Maas and Awni Y

    Andrew L. Maas and Awni Y. Hannun and Andrew Y. Ng , title =. ICML Workshop on Deep Learning for Audio, Speech and Language Processing , year =

  99. [111]

    2016 , url =

    Tongliang Liu and Dacheng Tao , title =. 2016 , url =. doi:10.1109/TPAMI.2015.2456899 , timestamp =

  100. [112]

    Reed and Honglak Lee and Dragomir Anguelov and Christian Szegedy and Dumitru Erhan and Andrew Rabinovich , editor =

    Scott E. Reed and Honglak Lee and Dragomir Anguelov and Christian Szegedy and Dumitru Erhan and Andrew Rabinovich , editor =. Training Deep Neural Networks on Noisy Labels with Bootstrapping , booktitle =. 2015 , url =

  101. [113]

    Hinton and Oriol Vinyals and Jeffrey Dean , title =

    Geoffrey E. Hinton and Oriol Vinyals and Jeffrey Dean , title =. CoRR , volume =. 2015 , url =. 1503.02531 , timestamp =

  102. [114]

    Mitchell , editor =

    Avrim Blum and Tom M. Mitchell , editor =. Combining Labeled and Unlabeled Data with Co-Training , booktitle =. 1998 , url =. doi:10.1145/279943.279962 , timestamp =

  103. [115]

    Dhillon and Pradeep Ravikumar and Ambuj Tewari , editor =

    Nagarajan Natarajan and Inderjit S. Dhillon and Pradeep Ravikumar and Ambuj Tewari , editor =. Learning with Noisy Labels , booktitle =. 2013 , url =

  104. [116]

    Manning and Andrew Y

    Richard Socher and Alex Perelygin and Jean Wu and Jason Chuang and Christopher D. Manning and Andrew Y. Ng and Christopher Potts , title =. Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing,. 2013 , url =

  105. [117]

    2009 , publisher=

    Learning multiple layers of features from tiny images , author=. 2009 , publisher=

  106. [118]

    Proceedings of the IEEE International Conference on Computer Vision , mendeley-groups =

    Li, Yuncheng and Yang, Jianchao and Song, Yale and Cao, Liangliang and Luo, Jiebo and Li, Li Jia , doi =. Proceedings of the IEEE International Conference on Computer Vision , mendeley-groups =. arXiv , arxivId =:1703.02391 , file =

  107. [119]

    ICML 2019 , mendeley-groups =

    Shen, Yanyao and Sanghavi, Sujay , eprint =. ICML 2019 , mendeley-groups =

  108. [120]

    6th International Conference on Learning Representations, ICLR 2018 - Conference Track Proceedings , mendeley-groups =

    Madry, Aleksander and Makelov, Aleksandar and Schmidt, Ludwig and Tsipras, Dimitris and Vladu, Adrian , eprint =. 6th International Conference on Learning Representations, ICLR 2018 - Conference Track Proceedings , mendeley-groups =

  109. [121]

    Kanwal and Tegan Maharaj and Asja Fischer and Aaron C

    Devansh Arpit and Stanislaw Jastrzebski and Nicolas Ballas and David Krueger and Emmanuel Bengio and Maxinder S. Kanwal and Tegan Maharaj and Asja Fischer and Aaron C. Courville and Yoshua Bengio and Simon Lacoste. A Closer Look at Memorization in Deep Networks , booktitle =. ...

  110. [122]

    Proceedings of the 37th International Conference on Machine Learning,

    Jiacheng Cheng and Tongliang Liu and Kotagiri Ramamohanarao and Dacheng Tao , title =. Proceedings of the 37th International Conference on Machine Learning,. 2020 , url =

  111. [123]

    9th International Conference on Learning Representations,

    Hao Cheng and Zhaowei Zhu and Xingyu Li and Yifei Gong and Xing Sun and Yang Liu , title =. 9th International Conference on Learning Representations,. 2021 , url =

  112. [124]

    Northcutt and Lu Jiang and Isaac L

    Curtis G. Northcutt and Lu Jiang and Isaac L. Chuang , title =. J. Artif. Intell. Res. , volume =. 2021 , url =. doi:10.1613/jair.1.12125 , timestamp =

  113. [125]

    Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition , mendeley-groups =

    Wang, Yisen and Liu, Weiyang and Ma, Xingjun and Bailey, James and Zha, Hongyuan and Song, Le and Xia, Shu Tao , doi =. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition , mendeley-groups =. arXiv , arxivId =:1804.00092 , file =

  114. [126]

    CoRR , volume =

    Sebastian Ruder , title =. CoRR , volume =. 2016 , url =. 1609.04747 , timestamp =

  115. [127]

    Deep Learning , author=

  116. [128]

    Proceedings - International Conference on Software Engineering , mendeley-groups =

    Wang, Jingyi and Chen, Jialuo and Sun, Youcheng and Ma, Xingjun and Wang, Dongxia and Sun, Jun and Cheng, Peng , doi =. Proceedings - International Conference on Software Engineering , mendeley-groups =. arXiv , arxivId =:2102.05913 , file =

  117. [129]

    Learning from Noisy Labels with Deep Neural Networks:

    Hwanjun Song and Minseok Kim and Dongmin Park and Jae. Learning from Noisy Labels with Deep Neural Networks:. CoRR , volume =. 2020 , url =. 2007.08199 , timestamp =

  118. [130]

    Giorgio Patrini and Alessandro Rozza and Aditya Krishna Menon and Richard Nock and Lizhen Qu , title =. 2017. 2017 , url =. doi:10.1109/CVPR.2017.240 , timestamp =

  119. [131]

    Understanding and Utilizing Deep Neural Networks Trained with Noisy Labels , booktitle =

    Pengfei Chen and Benben Liao and Guangyong Chen and Shengyu Zhang , editor =. Understanding and Utilizing Deep Neural Networks Trained with Noisy Labels , booktitle =. 2019 , url =

  120. [132]

    2019 , url =

    Hwanjun Song and Minseok Kim and Jae. 2019 , url =

  121. [134]

    and Sugiyama, Masashi , eprint =

    Yu, Xingrui and Han, Bo and Yao, Jiangchao and Niu, Gang and Tsang, Ivor W. and Sugiyama, Masashi , eprint =. 36th International Conference on Machine Learning, ICML 2019 , mendeley-groups =

  122. [135]

    35th International Conference on Machine Learning, ICML 2018 , keywords =

    Jiang, Lu and Zhou, Zhengyuan and Leung, Thomas and Li, Li Jia and Fei-Fei, Li , eprint =. 35th International Conference on Machine Learning, ICML 2018 , keywords =

  123. [136]

    Advances in Neural Information Processing Systems , mendeley-groups =

    Malach, Eran and Shalev-Shwartz, Shai , eprint =. Advances in Neural Information Processing Systems , mendeley-groups =

  124. [137]

    Robust Inference via Generative Classifiers for Handling Noisy Labels , booktitle =

    Kimin Lee and Sukmin Yun and Kibok Lee and Honglak Lee and Bo Li and Jinwoo Shin , editor =. Robust Inference via Generative Classifiers for Handling Noisy Labels , booktitle =. 2019 , url =

  125. [138]

    Are Anchor Points Really Indispensable in Label-Noise Learning? , booktitle =

    Xiaobo Xia and Tongliang Liu and Nannan Wang and Bo Han and Chen Gong and Gang Niu and Masashi Sugiyama , editor =. Are Anchor Points Really Indispensable in Label-Noise Learning? , booktitle =. 2019 , url =

  126. [139]

    and Sugiyama, Masashi , eprint =

    Han, Bo and Yao, Quanming and Yu, Xingrui and Niu, Gang and Xu, Miao and Hu, Weihua and Tsang, Ivor W. and Sugiyama, Masashi , eprint =. Advances in Neural Information Processing Systems , mendeley-groups =

  127. [140]

    Learning from Noisy Labels with Distillation , booktitle =

    Yuncheng Li and Jianchao Yang and Yale Song and Liangliang Cao and Jiebo Luo and Li. Learning from Noisy Labels with Distillation , booktitle =. 2017 , url =. doi:10.1109/ICCV.2017.211 , timestamp =

  128. [141]

    O'Connor and Kevin McGuinness , editor =

    Eric Arazo and Diego Ortego and Paul Albert and Noel E. O'Connor and Kevin McGuinness , editor =. Unsupervised Label Noise Modeling and Loss Correction , booktitle =. 2019 , url =

  129. [142]

    Active Bias: Training More Accurate Neural Networks by Emphasizing High Variance Samples , booktitle =

    Haw. Active Bias: Training More Accurate Neural Networks by Emphasizing High Variance Samples , booktitle =. 2017 , url =

  130. [143]

    Goodfellow and Jonathon Shlens and Christian Szegedy , title =

    Ian J. Goodfellow and Jonathon Shlens and Christian Szegedy , title =. 3rd International Conference on Learning Representations,. 2015 , url =

  131. [144]

    Training deep neural-networks using a noise adaptation layer , booktitle =

    Jacob Goldberger and Ehud Ben. Training deep neural-networks using a noise adaptation layer , booktitle =. 2017 , url =

  132. [145]

    Aritra Ghosh and Himanshu Kumar and P. S. Sastry , title =. Proceedings of the Thirty-First. 2017 , url =

  133. [146]

    2017 , url =

    Chiyuan Zhang and Samy Bengio and Moritz Hardt and Benjamin Recht and Oriol Vinyals , title =. 2017 , url =

  134. [147]

    doi:10.18653/v1/N19-1423 , eprint =

    Devlin, Jacob and Chang, Ming Wei and Lee, Kenton and Toutanova, Kristina , booktitle =. doi:10.18653/v1/N19-1423 , eprint =

  135. [148]

    Wong and Lidia S

    Qiang Wang and Bei Li and Tong Xiao and Jingbo Zhu and Changliang Li and Derek F. Wong and Lidia S. Chao , editor =. Learning Deep Transformer Models for Machine Translation , booktitle =. 2019 , url =. doi:10.18653/v1/p19-1176 , timestamp =

  136. [149]

    YOLOv4: Optimal Speed and Accuracy of Object Detection , journal =

    Alexey Bochkovskiy and Chien. YOLOv4: Optimal Speed and Accuracy of Object Detection , journal =. 2020 , url =. 2004.10934 , timestamp =

  137. [150]

    Marriott and Sami Romdhani and Liming Chen , title =

    Richard T. Marriott and Sami Romdhani and Liming Chen , title =. 2021 , url =

  138. [151]

    Visual Information Extraction with Lixto

    Robert Baumgartner and Georg Gottlob and Sergio Flesca. Visual Information Extraction with Lixto. Proceedings of the 27th International Conference on Very Large Databases. 2001

  139. [152]

    Brachman and James G

    Ronald J. Brachman and James G. Schmolze. An overview of the KL-ONE knowledge representation system. Cognitive Science. 1985

  140. [153]

    Complexity results for nonmonotonic logics

    Georg Gottlob. Complexity results for nonmonotonic logics. Journal of Logic and Computation. 1992

  141. [154]

    Hypertree Decompositions and Tractable Queries

    Georg Gottlob and Nicola Leone and Francesco Scarcello. Hypertree Decompositions and Tractable Queries. Journal of Computer and System Sciences. 2002

  142. [155]

    Levesque

    Hector J. Levesque. Foundations of a functional approach to knowledge representation. Artificial Intelligence. 1984

  143. [156]

    Levesque

    Hector J. Levesque. A logic of implicit and explicit belief. Proceedings of the Fourth National Conference on Artificial Intelligence. 1984

  144. [157]

    On the compilability and expressive power of propositional planning formalisms

    Bernhard Nebel. On the compilability and expressive power of propositional planning formalisms. Journal of Artificial Intelligence Research. 2000

  145. [158]

    Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , month =

    He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian , title =. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , month =

  146. [159]

    Identity Mappings in Deep Residual Networks , booktitle =

    Kaiming He and Xiangyu Zhang and Shaoqing Ren and Jian Sun , editor =. Identity Mappings in Deep Residual Networks , booktitle =. 2016 , url =. doi:10.1007/978-3-319-46493-0\_38 , timestamp =

  147. [160]

    Robust Training under Label Noise by Over-parameterization , booktitle =

    Sheng Liu and Zhihui Zhu and Qing Qu and Chong You , editor =. Robust Training under Label Noise by Over-parameterization , booktitle =. 2022 , url =

  148. [161]

    Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence,

    Ruixuan Xiao and Yiwen Dong and Haobo Wang and Lei Feng and Runze Wu and Gang Chen and Junbo Zhao , title =. Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence,. 2023 , url =. doi:10.24963/IJCAI.2023/494 , timestamp =

  149. [162]

    Sample Prior Guided Robust Model Learning to Suppress Noisy Labels , booktitle =

    Wenkai Chen and Chuang Zhu and Mengting Li , editor =. Sample Prior Guided Robust Model Learning to Suppress Noisy Labels , booktitle =. 2023 , url =. doi:10.1007/978-3-031-43415-0\_1 , timestamp =

  150. [163]

    CoRR , volume =

    Jiaheng Wei and Hangyu Liu and Tongliang Liu and Gang Niu and Yang Liu , title =. CoRR , volume =. 2021 , url =. 2106.04149 , timestamp =

  151. [164]

    Early-Learning Regularization Prevents Memorization of Noisy Labels , booktitle =

    Sheng Liu and Jonathan Niles. Early-Learning Regularization Prevents Memorization of Noisy Labels , booktitle =. 2020 , url =

  152. [165]

    Proceedings of the 37th International Conference on Machine Learning,

    Yang Liu and Hongyi Guo , title =. Proceedings of the 37th International Conference on Machine Learning,. 2020 , url =

  153. [166]

    Metaxas and Chao Chen , title =

    Songzhu Zheng and Pengxiang Wu and Aman Goswami and Mayank Goswami and Dimitris N. Metaxas and Chao Chen , title =. Proceedings of the 37th International Conference on Machine Learning,. 2020 , url =

  154. [167]

    Junnan Li and Richard Socher and Steven C. H. Hoi , title =. 8th International Conference on Learning Representations,. 2020 , url =

  155. [168]

    2020 , url =

    Duc Tam Nguyen and Chaithanya Kumar Mummadi and Thi. 2020 , url =

  156. [169]

    Thirty-Fifth

    Yichen Wu and Jun Shu and Qi Xie and Qian Zhao and Deyu Meng , title =. Thirty-Fifth. 2021 , url =

  157. [170]

    Kankanhalli , title =

    Junnan Li and Yongkang Wong and Qi Zhao and Mohan S. Kankanhalli , title =. 2019 , url =. doi:10.1109/CVPR.2019.00519 , timestamp =

  158. [171]

    CoRR , volume =

    Yuanpeng Tu and Boshen Zhang and Yuxi Li and Liang Liu and Jian Li and Yabiao Wang and Chengjie Wang and Cairong Zhao , title =. CoRR , volume =. 2023 , url =. doi:10.48550/arXiv.2302.06810 , eprinttype =. 2302.06810 , timestamp =

  159. [172]

    Proceedings of the 37th International Conference on Machine Learning , articleno =

    Ma, Xingjun and Huang, Hanxun and Wang, Yisen and Erfani, Simone Romano Sarah and Bailey, James , title =. Proceedings of the 37th International Conference on Machine Learning , articleno =. 2020 , publisher =

  160. [173]

    Dumais , title =

    Guoqing Zheng and Ahmed Hassan Awadallah and Susan T. Dumais , title =. Thirty-Fifth. 2021 , url =

  161. [174]

    2019 , url =

    Yisen Wang and Xingjun Ma and Zaiyi Chen and Yuan Luo and Jinfeng Yi and James Bailey , title =. 2019 , url =. doi:10.1109/ICCV.2019.00041 , timestamp =

  162. [175]

    Sabuncu , editor =

    Zhilu Zhang and Mert R. Sabuncu , editor =. Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labels , booktitle =. 2018 , url =

  163. [176]

    2019 , url =

    Kun Yi and Jianxin Wu , title =. 2019 , url =. doi:10.1109/CVPR.2019.00718 , timestamp =

  164. [178]

    Hinton , title =

    Ting Chen and Simon Kornblith and Mohammad Norouzi and Geoffrey E. Hinton , title =. Proceedings of the 37th International Conference on Machine Learning,. 2020 , url =

  165. [179]

    Supervised Contrastive Learning , booktitle =

    Prannay Khosla and Piotr Teterwak and Chen Wang and Aaron Sarna and Yonglong Tian and Phillip Isola and Aaron Maschinot and Ce Liu and Dilip Krishnan , editor =. Supervised Contrastive Learning , booktitle =. 2020 , url =

  166. [180]

    Representation Learning with Contrastive Predictive Coding , journal =

    A. Representation Learning with Contrastive Predictive Coding , journal =. 2018 , url =. 1807.03748 , timestamp =

  167. [181]

    CoRR , volume =

    Hansi Yang and Quanming Yao and Bo Han and Gang Niu , title =. CoRR , volume =. 2019 , url =. 1911.02377 , timestamp =

  168. [182]

    mixup: Beyond Empirical Risk Minimization , booktitle =

    Hongyi Zhang and Moustapha Ciss. mixup: Beyond Empirical Risk Minimization , booktitle =. 2018 , url =

  169. [183]

    2019 , url =

    Baoyun Peng and Xiao Jin and Dongsheng Li and Shunfeng Zhou and Yichao Wu and Jiaheng Liu and Zhaoning Zhang and Yu Liu , title =. 2019 , url =. doi:10.1109/ICCV.2019.00511 , timestamp =

  170. [184]

    Bilinear

    Tsung. Bilinear. 2015. 2015 , url =. doi:10.1109/ICCV.2015.170 , timestamp =

  171. [185]

    2022 , url =

    Jiaheng Wei and Zhaowei Zhu and Hao Cheng and Tongliang Liu and Gang Niu and Yang Liu , title =. 2022 , url =

  172. [186]

    and Long, Guodong and Yang, Yi , title =

    Yan, Yan and Xu, Zhongwen and Tsang, Ivor W. and Long, Guodong and Yang, Yi , title =. Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence , pages =. 2016 , publisher =

  173. [187]

    Goodfellow and Nicolas Papernot and Avital Oliver and Colin Raffel , editor =

    David Berthelot and Nicholas Carlini and Ian J. Goodfellow and Nicolas Papernot and Avital Oliver and Colin Raffel , editor =. MixMatch:. Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems 2019, NeurIPS 2019, Decemb...

  174. [188]

    Neural Information Processing Systems , year=

    Class-Dependent Label-Noise Learning with Cycle-Consistency Regularization , author=. Neural Information Processing Systems , year=

  175. [189]

    2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Learning with Neighbor Consistency for Noisy Labels , author=. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

  176. [190]

    2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    On Learning Contrastive Representations for Learning with Noisy Labels , author=. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

  177. [191]

    International Joint Conference on Artificial Intelligence , year=

    SELC: Self-Ensemble Label Correction Improves Learning with Noisy Labels , author=. International Joint Conference on Artificial Intelligence , year=

  178. [192]

    Neural Information Processing Systems , year=

    Noise Attention Learning: Enhancing Noise Robustness by Gradient Scaling , author=. Neural Information Processing Systems , year=

  179. [193]

    AAAI Conference on Artificial Intelligence , year=

    Beyond Class-Conditional Assumption: A Primary Attempt to Combat Instance-Dependent Label Noise , author=. AAAI Conference on Artificial Intelligence , year=

  180. [194]

    AAAI Conference on Artificial Intelligence Workshop , year=

    Model and Data Agreement for Learning with Noisy Labels , author=. AAAI Conference on Artificial Intelligence Workshop , year=

  181. [195]

    2023 , url=

    Mitigating Memorization of Noisy Labels by Clipping the Model Prediction , author=. 2023 , url=

  182. [196]

    The American Statistician , year=

    Thirteen ways to look at the correlation coefficient , author=. The American Statistician , year=

  183. [197]

    Two Wrongs Don't Make a Right: Combating Confirmation Bias in Learning with Label Noise , booktitle =

    Mingcai Chen and Hao Cheng and Yuntao Du and Ming Xu and Wenyu Jiang and Chongjun Wang , editor =. Two Wrongs Don't Make a Right: Combating Confirmation Bias in Learning with Label Noise , booktitle =. 2023 , url =. doi:10.1609/AAAI.V37I12.26725 , timestamp =

  184. [198]

    Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results , booktitle =

    Antti Tarvainen and Harri Valpola , editor =. Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results , booktitle =. 2017 , url =

  185. [199]

    Efficient Estimation of Word Representations in Vector Space , booktitle =

    Tom. Efficient Estimation of Word Representations in Vector Space , booktitle =. 2013 , url =

  186. [200]

    Contrastive Multiview Coding , booktitle =

    Yonglong Tian and Dilip Krishnan and Phillip Isola , editor =. Contrastive Multiview Coding , booktitle =. 2020 , url =. doi:10.1007/978-3-030-58621-8\_45 , timestamp =

  187. [201]

    Thirty-Sixth

    Chuanguang Yang and Zhulin An and Linhang Cai and Yongjun Xu , title =. Thirty-Sixth. 2022 , url =. doi:10.1609/AAAI.V36I3.20211 , timestamp =

  188. [202]

    2023 , url =

    Ziyi Zhang and Weikai Chen and Chaowei Fang and Zhen Li and Lechao Chen and Liang Lin and Guanbin Li , title =. 2023 , url =. doi:10.1109/ICCV51070.2023.00158 , timestamp =

  189. [203]

    CoRR , volume =

    Wen Li and Limin Wang and Wei Li and Eirikur Agustsson and Luc Van Gool , title =. CoRR , volume =. 2017 , url =. 1708.02862 , timestamp =

  190. [204]

    ImageNet:

    Jia Deng and Wei Dong and Richard Socher and Li. ImageNet:. 2009. 2009 , url =. doi:10.1109/CVPR.2009.5206848 , timestamp =

  191. [205]

    CrossSplit: Mitigating Label Noise Memorization through Data Splitting , booktitle =

    Jihye Kim and Aristide Baratin and Yan Zhang and Simon Lacoste. CrossSplit: Mitigating Label Noise Memorization through Data Splitting , booktitle =. 2023 , url =

  192. [207]

    2023 , url =

    Yuanpeng Tu and Boshen Zhang and Yuxi Li and Liang Liu and Jian Li and Yabiao Wang and Chengjie Wang and Cairong Zhao , title =. 2023 , url =. doi:10.1109/CVPR52729.2023.01909 , timestamp =

  193. [209]

    RandAugment: Practical Automated Data Augmentation with a Reduced Search Space , booktitle =

    Ekin Dogus Cubuk and Barret Zoph and Jonathon Shlens and Quoc Le , editor =. RandAugment: Practical Automated Data Augmentation with a Reduced Search Space , booktitle =. 2020 , url =

  194. [210]

    Taylor , title =

    Terrance Devries and Graham W. Taylor , title =. CoRR , volume =. 2017 , url =. 1708.04552 , timestamp =

  195. [211]

    2024 , url =

    Jingfeng Zhang and Bo Song and Haohan Wang and Bo Han and Tongliang Liu and Lei Liu and Masashi Sugiyama , title =. 2024 , url =. doi:10.1109/TPAMI.2024.3355425 , timestamp =

  196. [212]

    Tsang and Guodong Long and Yi Yang , editor =

    Yan Yan and Zhongwen Xu and Ivor W. Tsang and Guodong Long and Yi Yang , editor =. Robust Semi-Supervised Learning through Label Aggregation , booktitle =. 2016 , url =. doi:10.1609/AAAI.V30I1.10276 , timestamp =

  197. [213]

    Forty-first International Conference on Machine Learning,

    Jia Shi and Gautam Rajendrakumar Gare and Jinjin Tian and Siqi Chai and Zhiqiu Lin and Arun Balajee Vasudevan and Di Feng and Francesco Ferroni and Shu Kong , title =. Forty-first International Conference on Machine Learning,. 2024 , url =

  198. [214]

    Making Better Mistakes: Leveraging Class Hierarchies With Deep Networks , booktitle =

    Luca Bertinetto and Romain M. Making Better Mistakes: Leveraging Class Hierarchies With Deep Networks , booktitle =. 2020 , url =. doi:10.1109/CVPR42600.2020.01252 , timestamp =

  199. [215]

    2022 , url =

    Shikun Li and Xiaobo Xia and Shiming Ge and Tongliang Liu , title =. 2022 , url =. doi:10.1109/CVPR52688.2022.00041 , timestamp =

  200. [216]

    2015 , url =

    Tong Xiao and Tian Xia and Yi Yang and Chang Huang and Xiaogang Wang , title =. 2015 , url =. doi:10.1109/CVPR.2015.7298885 , timestamp =

  201. [217]

    Daehwan Kim and Kwangrok Ryoo and Hansang Cho and Seungryong Kim , title =. Int. J. Comput. Vis. , volume =. 2025 , url =. doi:10.1007/S11263-024-02187-4 , timestamp =

  202. [218]

    Instance-dependent label distribution estimation for learning with label noise

    Liao, Zehui and Hu, Shishuai and Xie, Yutong and Xia, Yong. Instance-dependent label distribution estimation for learning with label noise. Int. J. Comput. Vis

  203. [219]

    Variational rectification inference for learning with noisy labels

    Sun, Haoliang and Wei, Qi and Feng, Lei and Hu, Yupeng and Liu, Fan and Fan, Hehe and Yin, Yilong. Variational rectification inference for learning with noisy labels. Int. J. Comput. Vis

  204. [220]

    Mirco Planamente and Chiara Plizzari and Simone Alberto Peirone and Barbara Caputo and Andrea Bottino , title =. Int. J. Comput. Vis. , volume =. 2024 , url =. doi:10.1007/S11263-024-01998-9 , timestamp =

  205. [221]

    Mouxing Yang and Zhenyu Huang and Xi Peng , title =. Int. J. Comput. Vis. , volume =. 2024 , url =. doi:10.1007/S11263-024-01997-W , timestamp =

  206. [222]

    Using unreliable pseudo-labels for label-efficient semantic segmentation

    Wang, Haochen and Wang, Yuchao and Shen, Yujun and Fan, Junsong and Wang, Yuxi and Zhang, Zhaoxiang. Using unreliable pseudo-labels for label-efficient semantic segmentation. Int. J. Comput. Vis

  207. [223]

    When the Small-Loss Trick is Not Enough: Multi-Label Image Classification with Noisy Labels Applied to

    Keryan Chelouche and Marie Lachaize and Marine Bernard and Louise Olgiati and R. When the Small-Loss Trick is Not Enough: Multi-Label Image Classification with Noisy Labels Applied to. CoRR , volume =. 2024 , url =. doi:10.48550/ARXIV.2410.07689 , eprinttype =. 2410.07689 , ti...

  208. [224]

    Towards Understanding Deep Learning from Noisy Labels with Small-Loss Criterion , booktitle =

    Xian. Towards Understanding Deep Learning from Noisy Labels with Small-Loss Criterion , booktitle =. 2021 , url =. doi:10.24963/IJCAI.2021/340 , timestamp =

  209. [225]

    Efficient Approximation of Deep ReLU Networks for Functions on Low Dimensional Manifolds , booktitle =

    Minshuo Chen and Haoming Jiang and Wenjing Liao and Tuo Zhao , editor =. Efficient Approximation of Deep ReLU Networks for Functions on Low Dimensional Manifolds , booktitle =. 2019 , url =

  210. [226]

    CoRR , volume =

    Hao Chen and Jindong Wang and Zihan Wang and Ran Tao and Hongxin Wei and Xing Xie and Masashi Sugiyama and Bhiksha Raj , title =. CoRR , volume =. 2024 , url =. doi:10.48550/ARXIV.2403.06869 , eprinttype =. 2403.06869 , timestamp =

  211. [227]

    Can contrastive learning avoid shortcut solutions? , booktitle =

    Joshua Robinson and Li Sun and Ke Yu and Kayhan Batmanghelich and Stefanie Jegelka and Suvrit Sra , editor =. Can contrastive learning avoid shortcut solutions? , booktitle =. 2021 , url =

  212. [228]

    Shortcut learning in deep neural networks , journal =

    Robert Geirhos and J. Shortcut learning in deep neural networks , journal =. 2020 , url =. doi:10.1038/S42256-020-00257-Z , timestamp =

  213. [229]

    9th International Conference on Learning Representations,

    Yikai Zhang and Songzhu Zheng and Pengxiang Wu and Mayank Goswami and Chao Chen , title =. 9th International Conference on Learning Representations,. 2021 , url =

  214. [230]

    Yingyi Chen and Xi Shen and Shell Xu Hu and Johan A. K. Suykens , title =. 2021 , url =. doi:10.1109/CVPRW53098.2021.00302 , timestamp =

  215. [231]

    Combating Semantic Contamination in Learning with Label Noise , booktitle =

    Wenxiao Fan and Kan Li , editor =. Combating Semantic Contamination in Learning with Label Noise , booktitle =. 2025 , url =. doi:10.1609/AAAI.V39I3.32293 , timestamp =

  216. [232]

    Learning Disentangled Representations via Mutual Information Estimation , booktitle =

    Eduardo Hugo Sanchez and Mathieu Serrurier and Mathias Ortner , editor =. Learning Disentangled Representations via Mutual Information Estimation , booktitle =. 2020 , url =. doi:10.1007/978-3-030-58542-6\_13 , timestamp =

  217. [233]

    Thirty-Sixth

    Kai Guo and Kaixiong Zhou and Xia Hu and Yu Li and Yi Chang and Xin Wang , title =. Thirty-Sixth. 2022 , url =. doi:10.1609/AAAI.V36I4.20316 , timestamp =

  218. [234]

    Chao Yuan and Liming Yang , title =. Knowl. Based Syst. , volume =. 2022 , url =. doi:10.1016/J.KNOSYS.2022.108481 , timestamp =

  219. [235]

    M. G. Kendall , journal =. A New Measure of Rank Correlation , urldate =

  220. [236]

    The Rearrangement Inequality

    Cvetkovski, Zdravko. The Rearrangement Inequality. Inequalities: Theorems, Techniques and Selected Problems. 2012. doi:10.1007/978-3-642-23792-8_6

  221. [237]

    Bartlett and Dylan J

    Peter L. Bartlett and Dylan J. Foster and Matus Telgarsky , editor =. Spectrally-normalized margin bounds for neural networks , booktitle =. 2017 , url =

  222. [238]

    Generalization Bounds for Label Noise Stochastic Gradient Descent , booktitle =

    Jung Eun Huh and Patrick Rebeschini , editor =. Generalization Bounds for Label Noise Stochastic Gradient Descent , booktitle =. 2024 , url =

  223. [239]

    Cordeiro and Gustavo Carneiro , title =

    Filipe R. Cordeiro and Gustavo Carneiro , title =. Pattern Recognit. , volume =. 2025 , url =. doi:10.1016/J.PATCOG.2024.111132 , timestamp =

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.