Pith. sign in

REVIEW 4 major objections 6 minor 41 references

Robust Federated Learning with Confidence-Weighted Filtering and GAN-Based Completion under Noisy and Incomplete Data

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a three-stage federated learning pipeline—local confidence-weighted noise cleaning, federated conditional GAN training, and synthetic-data completion—makes federated classifiers robust to label noise, missing…

desk verdict A workmanlike pipeline of known components whose main empirical claim rests on a cleaning rule that, as written, does the opposite of what the text says. read the letter →

arxiv 2505.09733 v1 pith:UEKUWF4B submitted 2025-05-14 cs.LG cs.AIcs.CR

Alpaslan Gokcen , Ali Boyaci This is my paper · ORCID
classification cs.LGcs.AIcs.CR
keywords federatedlearninglabelnoiseconditionalGANdatacleaningnon-IIDclassimbalanceFedProxmacro-F1score
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that a three-stage federated learning pipeline can make standard aggregation rules robust to label noise, missing classes, and class imbalance without sharing raw client data. The first stage removes locally mislabeled samples using a confidence score built from prediction entropy, prediction margin, and silhouette-based cluster fit; the second trains a conditional GAN collaboratively; the third fills missing classes with synthetic samples before FedAvg or FedProx training. On MNIST and Fashion-MNIST with asymmetric label noise and missing classes, the cleaned variants are reported to beat noisy FedAvg and FedProx baselines by large macro-F1 margins, sometimes reaching 0.98. The paper's most consequential practical claim is that the cleaning stage alone, without GANs, provides most of the robustness gain, making the approach feasible on resource-constrained clients.

What carries the argument

The central object is the confidence filter's aggregate score in Algorithm 1: $C_{\mathrm{agg}}(x_i)=\frac{1}{3}(C_{\mathrm{ent}}(x_i)+C_{\mathrm{margin}}(x_i)+C_{\mathrm{cluster}}(x_i))$, with adaptive threshold $T=\frac{1}{3}(\mathrm{mean}(C)+\mathrm{median}(C)+P_{75}(C))$ and retention of samples with $C_{\mathrm{agg}}(x_i)\ge T$. In the paper's stated design, lower entropy, larger margin, and higher silhouette mark trustworthy samples, so this filter is supposed to discard mislabeled or ambiguous points before relabeling survivors with $\arg\max p$; the second mechanism is the federated conditional GAN, whose generator and discriminator weights are averaged with FedAvg each round and then used to create class-conditioned images for missing classes.

What would settle it

Run the cleaning stage on MNIST with 50% known label noise and compare the label-correctness rates of retained versus discarded samples; if retained samples are not clearly more often correct, or if the printed score formula preferentially keeps high-entropy samples, the reported gains cannot be attributed to the stated cleaning mechanism.

Watch

Extended reading notes

Core claim

The authors contend that data-quality corruption in federated settings should be handled before and during aggregation rather than by changing the classifier alone. They propose a modular pipeline: each client trains a small CNN with stratified K-fold cross-validation, computes per-sample confidence scores from prediction entropy, prediction margin, and silhouette-based cluster consistency, and removes samples below an adaptive threshold; a federated conditional GAN is then trained over the cleaned sets to synthesize samples for absent classes; finally, the completed datasets are used to train a global CNN with FedAvg or FedProx. The reported experiments show that the cleaning-only variants (CleanAvg, CleanProx) consistently beat the noisy FedAvg and FedProx baselines across 10–70% asymmetric label noise and 2–7 missing classes, with CleanProx reaching 0.98 macro-F1 on MNIST, and that GAN augmentation adds a smaller but visible gain on Fashion-MNIST, especially under severe imbalance.

Load-bearing premise

The load-bearing premise is that a locally trained CNN's prediction probabilities and silhouette scores separate mislabeled from correctly labeled samples—and, as written, the scoring formula adds raw entropy, so high-entropy samples score higher and may be retained instead of removed.

Editorial extensions

If this is right

  • Under the reported results, FedAvg and FedProx can recover most of their macro-F1 on MNIST by adding only the local cleaning stage, even with 70% label noise and seven missing classes.
  • On Fashion-MNIST, conditional-GAN augmentation gives a performance lift mainly when classes are severely missing, so generative completion is a targeted remedy for class sparsity rather than a universal add-on.
  • Because cleaning-only models match or beat GAN-augmented models on MNIST, the practical implication is that low-resource deployments can skip the GAN stage without losing much accuracy.
  • The framework keeps the communication pattern of standard FedAvg and FedProx, so its reported gains come from data-quality preprocessing plus the same round-based weight exchange.

Reading between the lines

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

  • Editorial: if Equation (1) is not a typo, the printed scoring formula adds raw entropy (which rises with uncertainty) to margin and silhouette, so the filter would preferentially retain high-entropy samples; the reported cleaning gains would then need another explanation, such as relabeling by argmax or shifts in class balance.
  • Editorial: the adaptive threshold based on mean, median, and 75th percentile depends on the score distribution, and under heavy noise a filter that keeps high-scoring samples may simply shrink the dataset; tracking retention ratios and whether retained samples are more often correctly labeled would separate cleaning from subsampling effects.
  • Editorial: because synthetic data quality is measured only through downstream F1, a direct quality check on generated images would clarify whether GAN completion adds genuine class coverage or acts mainly as a regularizer.
  • Editorial: the privacy guarantee asserted in the abstract is not measured in the experiments—the paper states that differential privacy mechanisms were not explicitly applied—so 'privacy compliant' should be read as architectural intent rather than a demonstrated property.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes a three-stage federated learning pipeline designed to mitigate label noise, missing classes, and class imbalance. Stage 1 is a client-level confidence-based noise cleaning step that computes entropy, margin, and silhouette scores for each sample, aggregates them, and filters out samples below an adaptive threshold. Stage 2 is a federated conditional GAN (cGAN) training procedure intended to generate synthetic samples for missing classes. Stage 3 completes local datasets with these synthetic samples and trains a global classifier with FedAvg or FedProx. The authors evaluate six model variants (with/without cleaning, with/without GAN augmentation, with FedAvg/FedProx) on MNIST and Fashion-MNIST under varying noise ratios and missing class sizes, reporting macro-F1 scores and claiming significant improvements over noisy baselines.

Significance. If the results were sound, the proposed modular framework would be a useful practical contribution to robust federated learning, particularly the idea of combining local filtering with synthetic data completion to address simultaneous data quality challenges. The paper provides clear algorithmic skeletons, a reasonable evaluation design across two datasets, and an ablation structure that isolates cleaning and augmentation effects. However, the central cleaning equation is internally inconsistent, the GAN training protocol is critically underspecified, and the reported numbers are presented without variance or significance information. As a result, the headline empirical claims are not currently backed by a technically coherent description of the method.

major comments (4)
  1. [III-B, Eq. (1), Algorithm 1] The aggregate confidence score in Eq. (1) is computed as the arithmetic mean of raw entropy, margin, and silhouette score. Raw entropy is largest for the most uncertain predictions, so high-entropy samples receive high aggregate scores and are retained by the C_agg ≥ T criterion, while confident samples are more likely to be discarded. This is the opposite of the stated goal of "retaining high quality data" and "isolating and removing mislabeled instances," and the threshold in Eq. (2) does not repair the sign inversion. Because the cleaning stage is the main source of the reported gains in Tables II and V, this is a load-bearing error: the implementation either contradicts the written algorithm or the described mechanism would not produce the claimed improvements.
  2. [III-C, Algorithm 2] The federated GAN training stage never defines the adversarial objective. Algorithm 2 repeatedly calls a function train_one_epoch(G_global, D_global, μ) and averages the returned weights, but the paper does not specify the generator loss, the discriminator loss, the conditioning mechanism (how the label y is fed to the generator and discriminator beyond a concatenation mentioned in Section IV), or the gradient updates. Weight averaging alone does not define a GAN. This makes the "collaborative conditional GAN" unimplementable and unreproducible, and it undermines all claims about synthetic data quality in Section V-C.
  3. [IV-B, Algorithm 5] The text in Section IV-B states that asymmetric label noise is introduced by "assigning them to semantically related but incorrect classes," but Algorithm 5 assigns new labels by "sampling from the label distribution of the valid set," which is a uniform/random relabeling among the available classes and is not semantically related. Since the experimental results depend on which noise model is actually used, this mismatch must be resolved and the experiments described accordingly.
  4. [V, Table II] The paper states in Section IV that each experiment was repeated 50 times and average results were reported, yet Table II and the figures present only point estimates with no standard deviations, confidence intervals, or significance tests. Given that the central claim is "significant improvements in federated model performance," the lack of any variance information makes it impossible to assess whether the differences between models are meaningful, especially for close comparisons such as CleanAvg versus GenCleanAvg.
minor comments (6)
  1. [Abstract / Section VII] The abstract and conclusion claim the framework "rigorously maintains data privacy" and "incorporates differential privacy mechanisms," but Section VII states that differential privacy mechanisms were not explicitly applied in the current experiments. These statements should be reconciled.
  2. [III-A / IV-C] The paper mentions "confidence weighted aggregation" in the introduction and related work, but Algorithm 4 aggregates client updates by sample count only. No mechanism for weighting updates by confidence scores is described, so the terminology is misleading.
  3. [IV-B] The noise injection equation in Algorithm 5 has two branches that appear to be labeled inconsistently with the surrounding text; please clarify when each branch is used and how the final noise ratio is guaranteed for small client datasets.
  4. [Table II] The formatting of Table II runs model names and numeric values together, making it very hard to read. Clear column separators or a per-dataset layout would be necessary for a journal publication.
  5. [II, reference [28]] The citation for entropy-based confidence is about knowledge entropy decay during language model pretraining, which is not the standard reference for prediction confidence or calibration. Please cite a standard source on softmax entropy or confidence estimation.
  6. [Section V-B] The paper does not report the number of synthetic samples added per missing class in the actual experiments, although Algorithm 3 has a parameter s. Without this information, the effect of the GAN augmentation cannot be quantitatively interpreted.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported gains are evaluated against external test benchmarks and no load-bearing step reduces to its own inputs.

full rationale

The paper's central derivation chain is empirical rather than circular. Noise cleaning in Algorithm 1 filters and relabels local samples using out-of-fold CNN predictions and an adaptive threshold, but the final macro-F1 scores are computed on held-out MNIST and Fashion-MNIST test sets that are not used in cleaning, thresholding, or GAN training. No fitted parameter is renamed as a prediction: the threshold in Eq. (2) is computed from the score distribution, not optimized against the reported test metric, and the reported improvements are not forced by construction. The only self-citation, reference [5], is used merely to motivate federated learning in connected vehicles and plays no role in the method or evaluation. The concern that Eq. (1) adds raw entropy, which is large for uncertain samples, into an aggregate 'confidence' score is a potential correctness or sign bug, not a circular reduction: it does not make the reported F1 values true by definition. The GAN completion stage is also evaluated only through downstream classification on external test data, and the paper explicitly acknowledges that FID/IS metrics were not used and that GAN gains are dataset-dependent. Overall, no derivation in the paper is equivalent to its inputs, so the circularity burden is not met.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the cleaning threshold rule, the unspecified GAN objective, and the assumption that synthetic samples are trustworthy. The paper does not report several critical hyperparameters. No genuinely new conceptual entities are introduced.

free parameters (5)
  • Adaptive threshold statistic mixture = (mean + median + P75)/3
    Equation (2) defines the cleaning threshold as the average of three statistics; this exact combination is chosen by hand, no sensitivity analysis is given, and it directly controls the amount of data removed.
  • FedProx proximal coefficient mu = 0.01
    Set from the FedProx paper, not tuned or ablated in this study.
  • Synthetic samples per class s
    Algorithm 3 requires a per-class sample size s, but the value used in experiments is never reported; it controls how much class imbalance is corrected.
  • Number of cross-validation folds K
    Algorithm 1 uses K stratified folds; K is not reported, though it affects which samples are flagged as noisy.
  • K-means cluster count for cluster confidence
    Cluster-based confidence uses K-means on features plus probabilities; the cluster count is not specified.
assumptions (4)
  • domain assumption Local CNN classifiers can serve as reliable confidence estimators for label noise
    Stage 1 assumes a small CNN trained on noisy local data produces probabilities that separate clean from noisy samples; Section III B.
  • ad hoc to paper The aggregate confidence score in Equation (1) ranks sample trustworthiness as intended
    The mixture rule and particularly the sign of the entropy term are assumed valid, but the text contains an internal contradiction about whether high entropy lowers or raises the score.
  • ad hoc to paper Synthetic samples generated by the federated cGAN are label-faithful and class-representative
    Stage 3 relies on generated samples being semantically correct for missing classes; only indirect downstream F1 is reported, no FID/IS or manual inspection.
  • standard math FedAvg and FedProx converge as usual in the presence of generated data
    The paper uses standard FL aggregation without providing convergence analysis for the combined pipeline.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Robust Federated Learning with Confidence-Weighted Filtering and GAN-Based Completion under Noisy and Incomplete Data." pith.science (2026). https://pith.science/paper/UEKUWF4B

@misc{pith2026250509733,
  author       = {Pith},
  title        = {Pith review of: Robust Federated Learning with Confidence-Weighted Filtering and GAN-Based Completion under Noisy and Incomplete Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UEKUWF4B}},
  note         = {Machine review of arXiv:2505.09733}
}
read the original abstract

Federated learning (FL) presents an effective solution for collaborative model training while maintaining data privacy across decentralized client datasets. However, data quality issues such as noisy labels, missing classes, and imbalanced distributions significantly challenge its effectiveness. This study proposes a federated learning methodology that systematically addresses data quality issues, including noise, class imbalance, and missing labels. The proposed approach systematically enhances data integrity through adaptive noise cleaning, collaborative conditional GAN-based synthetic data generation, and robust federated model training. Experimental evaluations conducted on benchmark datasets (MNIST and Fashion-MNIST) demonstrate significant improvements in federated model performance, particularly macro-F1 Score, under varying noise and class imbalance conditions. Additionally, the proposed framework carefully balances computational feasibility and substantial performance gains, ensuring practicality for resource constrained edge devices while rigorously maintaining data privacy. Our results indicate that this method effectively mitigates common data quality challenges, providing a robust, scalable, and privacy compliant solution suitable for diverse real-world federated learning scenarios.

Figures

Figures reproduced from arXiv: 2505.09733 by the authors.

Figure 1
Figure 1. Three-Stage Federated Learning Framework for Robust Training under [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. F1 Score across varying noise ratios for models trained on MNIST [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. F1 Score versus number of missing classes across models and datasets. [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Heatmap of average F1 scores for each model on MNIST and [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 23 canonical work pages

  1. [5]

    Privacy-preserving real-time action detection in intelligent vehicles using federated learning-based temporal recurrent network,

    A. G ¨okcen and A. Boyacı, “Privacy-preserving real-time action detection in intelligent vehicles using federated learning-based temporal recurrent network,”Electronics, vol. 13, no. 14, 2024. [Online]. Available: https://www.mdpi.com/2079-9292/13/14/2820

    work page 2024

  2. [1]

    Communication-efficient learning of deep networks from decentralized data,

    H. B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, “Communication-efficient learning of deep networks from decentralized data,” inInternational Conference on Artificial Intelligence and Statistics, 2016. [Online]. Available: https://api.semanticscholar.org/ CorpusID:14955348

    work page 2016

  3. [2]

    Federated machine learning in healthcare: A systematic review on clinical applications and technical architecture,

    Z. L. Teo, L. Jin, N. Liu, S. Li, D. Miao, X. Zhang, W. Y . Ng, T. F. Tan, D. M. Lee, K. J. Chua, J. Heng, Y . Liu, R. S. M. Goh, and D. S. W. Ting, “Federated machine learning in healthcare: A systematic review on clinical applications and technical architecture,”Cell Reports Medicine, vol. 5, no. 2, p. 101419, Feb. 2024. [Online]. Available: http://dx.d...

    arXiv 2024

  4. [3]

    Efficient and secure federated learning for financial applications,

    T. Liu, Z. Wang, H. He, W. Shi, L. Lin, W. Shi, R. An, and C. Li, “Efficient and secure federated learning for financial applications,”

    work page

  5. [4]

    Federated learning in mobile edge computing: An edge-learning perspective for beyond 5g,

    S. Jere, Q. Fan, B. Shang, L. Li, and L. Liu, “Federated learning in mobile edge computing: An edge-learning perspective for beyond 5g,”

    work page

  6. [6]

    A survey of federated learning for edge computing: Research problems and solutions,

    Q. Xia, W. Ye, Z. Tao, J. Wu, and Q. Li, “A survey of federated learning for edge computing: Research problems and solutions,”High-Confidence Computing, vol. 1, no. 1, p. 100008, 2021. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S266729522100009X

    work page 2021

  7. [7]

    Federated learning: Challenges, methods, and future directions,

    T. Li, A. K. Sahu, A. Talwalkar, and V . Smith, “Federated learning: Challenges, methods, and future directions,”IEEE Signal Processing Magazine, vol. 37, no. 3, p. 50–60, May 2020. [Online]. Available: http://dx.doi.org/10.1109/MSP.2020.2975749

    arXiv 2020

  8. [8]

    Issues in federated learning: some experiments and preliminary results,

    J. Bhanbhro, S. Nistic `o, and L. Palopoli, “Issues in federated learning: some experiments and preliminary results,”Scientific Reports, vol. 14, no. 1, Dec. 2024. [Online]. Available: http: //dx.doi.org/10.1038/s41598-024-81732-0

    doi:10.1038/s41598-024-81732-0 2024

Show all 41 references
  1. [9]

    Mnist handwritten digit database,

    Y . LeCun, C. Cortes, and C. Burges, “Mnist handwritten digit database,” ATT Labs [Online]. Available: http://yann.lecun.com/exdb/mnist, vol. 2, 2010

    2010

  2. [10]

    Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms,

    H. Xiao, K. Rasul, and R. V ollgraf, “Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms,”CoRR, vol. abs/1708.07747, 2017. [Online]. Available: http://arxiv.org/abs/1708. 07747

    2017 arXiv

  3. [11]

    Federated learning with non-iid data,

    Y . Zhao, M. Li, L. Lai, N. Suda, D. Civin, and V . Chandra, “Federated learning with non-iid data,” 2018. [Online]. Available: https://arxiv.org/abs/1806.00582

    2018 arXiv

  4. [12]

    Generative models for effective ml on private, decentralized datasets,

    S. Augenstein, H. B. McMahan, D. Ramage, S. Ramaswamy, P. Kairouz, M. Chen, R. Mathews, and B. A. y Arcas, “Generative models for effective ml on private, decentralized datasets,” 2020. [Online]. Available: https://arxiv.org/abs/1911.06679

    2020 arXiv

  5. [13]

    Robust federated learning with noisy labels,

    S. Yang, H. Park, J. Byun, and C. Kim, “Robust federated learning with noisy labels,”IEEE Intelligent Systems, vol. 37, no. 2, pp. 35–43, 2022

    2022

  6. [14]

    Fedcg: Leverage conditional gan for protecting privacy and maintaining competitive performance in federated learning,

    Y . Wu, Y . Kang, J. Luo, Y . He, L. Fan, R. Pan, and Q. Yang, “Fedcg: Leverage conditional gan for protecting privacy and maintaining competitive performance in federated learning,” inProceedings of the Thirty-First International Joint Conference on Artificial Intelligence, s...

    2022 doi

  7. [15]

    Fedar+: A federated learning approach to appliance recognition with mislabeled data in residential environments,

    A. Gupta, H. P. Gupta, and S. K. Das, “Fedar+: A federated learning approach to appliance recognition with mislabeled data in residential environments,” inProceedings of the ACM/IEEE 14th International Conference on Cyber-Physical Systems (with CPS-IoT Week 2023), ser. ICCPS ’...

    2023

  8. [16]

    Learning cautiously in federated learning with noisy and heterogeneous clients,

    C. Wu, Z. Li, F. Wang, and C. Wu, “Learning cautiously in federated learning with noisy and heterogeneous clients,” in2023 IEEE Interna- tional Conference on Multimedia and Expo (ICME), 2023, pp. 660–665

    2023

  9. [17]

    Fednoro: towards noise-robust federated learning by addressing class imbalance and label noise heterogeneity,

    N. Wu, L. Yu, X. Jiang, K.-T. Cheng, and Z. Yan, “Fednoro: towards noise-robust federated learning by addressing class imbalance and label noise heterogeneity,” inProceedings of the Thirty-Second International Joint Conference on Artificial Intelligence, ser. IJCAI ’23, 2023. ...

    2023 doi

  10. [18]

    Fednoisy: Federated noisy label learning benchmark,

    S. Liang, J. Huang, J. Hong, D. Zeng, J. Zhou, and Z. Xu, “Fednoisy: Federated noisy label learning benchmark,” 2025. [Online]. Available: https://arxiv.org/abs/2306.11650

    2025 arXiv

  11. [19]

    Collaboratively learning federated models from noisy decentralized data,

    H. Li, M. Funk, N. M. G ¨urel, and A. Saeed, “Collaboratively learning federated models from noisy decentralized data,” in2024 IEEE Interna- tional Conference on Big Data (BigData). IEEE, 2024, pp. 7879–7888

    2024

  12. [20]

    Federated learning client pruning for noisy labels,

    M. Morafah, H. Chang, C. Chen, and B. Lin, “Federated learning client pruning for noisy labels,”ACM Trans. Model. Perform. Eval. Comput. Syst., Nov. 2024, just Accepted. [Online]. Available: https://doi.org/10.1145/3706058

    2024 doi

  13. [21]

    Federated learning for automatic modulation classification under class imbalance and varying noise condition,

    Y . Wang, G. Gui, H. Gacanin, B. Adebisi, H. Sari, and F. Adachi, “Federated learning for automatic modulation classification under class imbalance and varying noise condition,”IEEE Transactions on Cognitive Communications and Networking, vol. 8, no. 1, pp. 86–96, 2022

    2022

  14. [22]

    Robust federated learning with noisy and hetero- geneous clients,

    X. Fang and M. Ye, “Robust federated learning with noisy and hetero- geneous clients,” in2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022, pp. 10 062–10 071

    2022

  15. [23]

    Federated semi- supervised learning with inter-client consistency & disjoint learning,

    W. Jeong, J. Yoon, E. Yang, and S. J. Hwang, “Federated semi- supervised learning with inter-client consistency & disjoint learning,” in International Conference on Learning Representations, 2021. [Online]. Available: https://openreview.net/forum?id=ce6CFXBh30h

    2021

  16. [24]

    Federated learning with client-exclusive classes,

    J. Zhang, X. Zhang, X. Zhang, D. Hong, R. Gupta, and J. Shang, “Federated learning with client-exclusive classes,” 01 2023

    2023

  17. [25]

    Federated optimization in heterogeneous networks,

    T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, and V . Smith, “Federated optimization in heterogeneous networks,” 2020. [Online]. Available: https://arxiv.org/abs/1812.06127

    2020 arXiv

  18. [26]

    Review of deep learning: concepts, cnn architectures, challenges, applications, future directions,

    L. Alzubaidi, J. Zhang, A. J. Humaidi, A. Al-Dujaili, Y . Duan, O. Al-Shamma, J. Santamar ´ıa, M. A. Fadhel, M. Al-Amidie, and L. Farhan, “Review of deep learning: concepts, cnn architectures, challenges, applications, future directions,”Journal of Big Data, vol. 8, no. 1, Mar...

    2021

  19. [27]

    A study of cross-validation and bootstrap for accuracy estimation and model selection,

    R. Kohavi, “A study of cross-validation and bootstrap for accuracy estimation and model selection,” inInternational Joint Conference on Artificial Intelligence, 1995. [Online]. Available: https://api. semanticscholar.org/CorpusID:2702042

    1995

  20. [28]

    Knowledge entropy decay during language model pretraining hinders new knowledge acquisition,

    J. Kim, H. Lee, H. Cho, J. Jang, H. Hwang, S. Won, Y . Ahn, D. Lee, and M. Seo, “Knowledge entropy decay during language model pretraining hinders new knowledge acquisition,” 2025. [Online]. Available: https://arxiv.org/abs/2410.01380

    2025 arXiv

  21. [29]

    Some methods for classification and analysis of multivariate observations,

    J. MacQueen, “Some methods for classification and analysis of multivariate observations,” 1967. [Online]. Available: https://api. semanticscholar.org/CorpusID:6278891

    1967

  22. [30]

    Least squares quantization in pcm,

    S. P. Lloyd, “Least squares quantization in pcm,”IEEE Trans. Inf. Theory, vol. 28, pp. 129–136, 1982. [Online]. Available: https://api.semanticscholar.org/CorpusID:10833328

    1982

  23. [31]

    Conditional generative adversarial nets,

    M. Mirza and S. Osindero, “Conditional generative adversarial nets,”

  24. [32]

    Gradient-based learning applied to document recognition,

    Y . Lecun, L. Bottou, Y . Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,”Proceedings of the IEEE, vol. 86, pp. 2278 – 2324, 12 1998

    1998

  25. [33]

    Dropout: A simple way to prevent neural networks from overfit- ting,

    N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, and R. Salakhut- dinov, “Dropout: A simple way to prevent neural networks from overfit- ting,”Journal of Machine Learning Research, vol. 15, pp. 1929–1958, 06 2014

    1929

  26. [34]

    Batch normalization: Accelerating deep network training by reducing internal covariate shift,

    S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network training by reducing internal covariate shift,” 2015. [Online]. Available: https://arxiv.org/abs/1502.03167

    2015 arXiv

  27. [35]

    Unsupervised representation learning with deep convolutional generative adversarial networks,

    A. Radford, L. Metz, and S. Chintala, “Unsupervised representation learning with deep convolutional generative adversarial networks,”

  28. [36]

    Evaluating the suitability of inception score and fr ´echet inception distance as metrics for quality and diversity in image generation,

    D. A. Chan and S. P. Sithungu, “Evaluating the suitability of inception score and fr ´echet inception distance as metrics for quality and diversity in image generation,” inProceedings of the 2024 7th International Conference on Computational Intelligence and Intelligent System...

    2024

  29. [37]

    Improved techniques for training gans,

    T. Salimans, I. Goodfellow, W. Zaremba, V . Cheung, A. Radford, X. Chen, and X. Chen, “Improved techniques for training gans,” inAdvances in Neural Information Processing Systems, D. Lee, M. Sugiyama, U. Luxburg, I. Guyon, and R. Garnett, Eds., vol. 29. Curran Associates, Inc....

    2016

  30. [2014]

    Available: https://arxiv.org/abs/1411.1784

    [Online]. Available: https://arxiv.org/abs/1411.1784

    arXiv

  31. [2016]

    Available: https://arxiv.org/abs/1511.06434

    [Online]. Available: https://arxiv.org/abs/1511.06434

    arXiv

  32. [2020]

    Available: https://arxiv.org/abs/2007.08030

    [Online]. Available: https://arxiv.org/abs/2007.08030

    2007 arXiv

  33. [2023]

    Available: https://arxiv.org/abs/2303.08355

    [Online]. Available: https://arxiv.org/abs/2303.08355

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.