Pith. sign in

REVIEW 3 major objections 6 minor 27 references

Federated Domain Generalization with Label Smoothing and Balanced Decentralized Training

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read FedSB applies client-level label smoothing and a fixed per-client training budget to federated domain generalization, reporting state-of-the-art accuracy on three of four standard domain-shift benchmarks.

desk verdict FedSB is a plausible, simple FDG method whose SOTA claim depends on an undocumented hyperparameter selection protocol. read the letter →

arxiv 2412.11408 v2 pith:4YVE7DL3 submitted 2024-12-16 cs.LG cs.AI

Milad Soltany , Farhad Pourpanah , Mahdiyar Molahasani , Michael Greenspan , Ali Etemad This is my paper · ORCID
classification cs.LGcs.AI
keywords federatedlearningdomaingeneralizationlabelsmoothingdataheterogeneityclientbudgetingdecentralizedtrainingshift
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

FedSB is a federated learning method for domain generalization built from two ordinary-looking changes to standard federated averaging. Each client trains with label-smoothed targets, replacing hard one-hot labels with a mixture that keeps most probability on the true class and spreads the rest uniformly, which the authors argue reduces local overconfidence and prevents overfitting to domain-specific features. In addition, every client is forced to train on the same number of samples per round by randomly subsampling large local datasets and oversampling small ones, so no client dominates the global model merely because it has more data. The paper reports that this combination achieves state-of-the-art average accuracy on three of the four DomainBed benchmarks used (PACS, OfficeHome, and TerraIncognita), with an ablation on PACS showing each component helps. If the result holds, two simple and privacy-preserving tweaks capture much of the benefit that more elaborate federated domain generalization methods buy with extra communication or shared statistics.

What carries the argument

The load-bearing object is the pair formed by the client-level label-smoothing loss and the fixed training budget $S$. Label smoothing replaces the one-hot target with $y'_c = 1-\epsilon+\epsilon/M$ for the true class and $\epsilon/M$ elsewhere; expanded, this makes the loss a weighted sum of the negative log-likelihood on the true class and a smooth loss that penalizes confident wrong-class probabilities. The budget mechanism resamples each local dataset to exactly $S$ samples before local training, which in Eqs. (7)--(8) removes the $|\mathcal{D}_i|$ factor from the expected global update. What the pair does is carry the causal story: local overconfidence and dataset-size imbalance are two separate, addressable mechanisms that hurt the aggregated model, and each can be fixed without changing the federated communication protocol.

What would settle it

Construct a two-client version of PACS in which the small-client domain has deliberately noisy labels while the large-client domain is clean, and compare FedSB under budgets $S$ equal to the small and large dataset sizes. If the fixed-budget rule in Eq. (8) is correct, equalizing the noisy small client should not hurt; observing a large accuracy drop when the noisy small client is force-equalized would falsify the equal-contribution premise.

Watch

Extended reading notes

Core claim

FedSB's central claim is that overconfident local models and unequal client data volumes are the two main correctable causes of poor cross-domain generalization in federated learning. Label smoothing changes the local loss to $(1-\epsilon)(-\log p_y) + \epsilon(-\frac{1}{M}\sum_c \log p_c)$, so a client is rewarded for calibrated, less certain predictions rather than for locking onto its own domain's shortcuts. The budgeting mechanism resamples each local dataset $\mathcal{D}_i$ to a fixed size $S$, and Eq. (8) then makes the expected global update equal to $\Theta^t - \frac{\eta S}{KB}\sum_i \mathbb{E}[\nabla \hat{\mathcal{L}}_i]$, independent of $|\mathcal{D}_i|$. With both components, the paper reports averages of 83.81 on PACS, 70.92 on OfficeHome, 46.89 on TerraIncognita, and 76.64 on VLCS, and its ablation on PACS shows that removing either the smoothing or the budget lowers the average accuracy.

Load-bearing premise

The balancing mechanism assumes that sample count, rather than domain difficulty, class imbalance, or label noise, is the main source of client influence, so that randomly shrinking large clients and oversampling small clients to the same size $S$ equalizes contributions without introducing harmful distortion.

Editorial extensions

If this is right

  • Client-level label smoothing can be dropped into any federated averaging scheme without sharing data or changing communication, so it is a cheap plug-in for improving unseen-domain accuracy.
  • Equalizing training budgets removes the dataset-size term from the expected global update, so sample-rich clients no longer steer the model toward their own domain.
  • The reported state-of-the-art averages on PACS (83.81), OfficeHome (70.92), and TerraIncognita (46.89) suggest the method transfers across very different domain-shift types.
  • Ablations on PACS indicate both ingredients matter: smoothing alone gives 83.31, budgeting alone gives 81.51, and the full method gives 83.81.
  • The method also improves over FedAvg with ViT backbones (for example, PACS ViT-b/16 average 88.56 versus 87.24), so the benefit is not tied to ResNet architectures.

Reading between the lines

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

  • Because the budget step only resamples data and does not alter the aggregation rule, FedSB's balancing idea should combine with FedProx-style proximal terms or prototype-sharing methods, but the paper does not test those combinations.
  • A natural check the paper does not run is to make the largest client the cleanest and easiest domain; if equalizing budgets still helps, the benefit is not simply removing a dominant domain, and if it hurts, the equal-contribution premise needs qualification.
  • The label-smoothing component could in principle be applied to client-level objectives in non-federated domain generalization as well, where the same overconfidence problem appears, but that transfer is outside the paper's experiments.
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

3 major / 6 minor

Summary. The manuscript proposes FedSB, a federated domain generalization method that combines client-level label smoothing with a per-client training budget. Label smoothing replaces hard one-hot labels with soft labels to reduce client overconfidence, while the budget mechanism resamples every local dataset to a fixed size S by under-sampling large clients and oversampling small clients. The authors argue, via Eq. (8), that this ensures equal contribution of clients to the global update. They evaluate FedSB on PACS, OfficeHome, TerraIncognita, and VLCS under the leave-one-domain-out protocol, reporting state-of-the-art accuracy on three of the four datasets, and include ablations on PACS showing that both components improve accuracy over FedAvg. Additional experiments use ResNet-50 and Vision Transformer backbones, and the code is publicly released.

Significance. If the empirical claims hold, FedSB is a simple and inexpensive addition to federated domain generalization: it requires no extra communication, no style sharing, and no auxiliary networks, only a resampling step and a modified loss. The manuscript is clearly written and gives credit-worthy practical details: ablations isolate the contribution of each component, sensitivity analysis is provided, multiple backbones are tested, and the code is released. The central risk is not the internal consistency of the derivation but the external validity of the benchmark numbers: the hyperparameter selection protocol is not described, and the reported margins over the strongest baseline are small enough that test-set-guided selection of epsilon and S could account for them. The contribution is therefore promising but not yet fully substantiated.

major comments (3)
  1. [IV (Implementation details), Tables I and VI] The manuscript never states how the label-smoothing coefficient epsilon and the per-client budget S are chosen, nor does it report the final values used for each dataset in Table I. Table VI shows that varying S alone changes PACS average accuracy from 80.27 to 81.51, and varying epsilon changes it from 82.98 to 83.35, while FedSB's reported margin over FedIIR on PACS is only 1.45 points. Without a stated model-selection rule (e.g., the training-domain validation split used in the DomainBed protocol), the claimed state-of-the-art results could be the product of test-domain-guided hyperparameter tuning. Please specify the selection protocol and report the per-dataset values of epsilon and S, including the meaning of the 'B' unit in entries such as S=45B.
  2. [III, Eq. (8)] Equation (8) shows that after resampling every local dataset to S samples, the expected global update is an average over K clients of per-client gradient sums, with a common factor S/(KB). This establishes equal sample counts, but not equal influence on the learned model: random subsampling of large clients may alter class balance, and oversampling small clients duplicates samples and can amplify label noise. The paper does not analyze these effects, so the statement that the budget 'ensures equal contribution from all clients' is stronger than what the derivation supports. The ablation in Table V indicates an empirical benefit on PACS, but the mechanistic claim needs qualification or additional analysis.
  3. [IV, Baselines paragraph] FedADG is the only baseline not re-run by the authors; its numbers in Table I are taken from the original paper. If FedADG was trained with a different backbone, learning-rate schedule, or number of communication rounds, the comparison in Table I is not apples-to-apples. The authors should either re-run FedADG under the same protocol or provide a clear statement of the differences and justify that the comparison remains fair.
minor comments (6)
  1. [Tables II and III] The header 'FebSB' in Tables II and III should be 'FedSB'.
  2. [Table I] The numbers in Table I are concatenated without spacing (e.g., '91.67±0.979.25±1.3...'), making the table difficult to parse; please format the entries with proper separators.
  3. [III, after Eq. (8)] The phrase 'were ˆL denotes' should be 'where ˆL denotes'.
  4. [III, Section III and Table VI] The text defines S as a number of samples, but Table VI uses units such as '30B' and '45B', suggesting batches; please clarify the relation between S and the batch size used in training.
  5. [Throughout] The dataset name is written as 'TerraINC' in Table I and as 'TerraIncognita' elsewhere; please standardize the notation.
  6. [IV, Implementation details] The paper says 'Following [11], we utilize the leave-one-domain-out setting' but does not mention whether the DomainBed training-domain validation rule for model selection is followed; even if a different rule is used, it should be stated explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: FedSB's derivation is self-contained; the balanced-budget equation is a definitional identity of the resampling procedure, not a fitted prediction.

full rationale

FedSB's derivation chain is self-contained. Label smoothing is introduced via Eq. (1), and the loss decomposition in Eq. (3) is a standard algebraic rearrangement of cross-entropy with smoothed labels; it is not used to derive the empirical accuracy gains. The balanced-training argument in Eq. (8) follows by construction from resampling every client's local dataset to a fixed size S, so the statement that all clients contribute equally to the expected global update is an identity of the resampling procedure rather than a prediction fitted to data. The claimed state-of-the-art results rest on experimental comparisons, ablations, and sensitivity analyses, and no parameter is fit to the target results. The self-citation in the related work, reference [24], describes a separate unsupervised FDG method and is not load-bearing for FedSB's mechanism. The lack of a stated hyperparameter-selection protocol is a reproducibility and comparison-protocol concern, not circularity. Therefore no circular step is exhibited.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

No new physical or conceptual entities are introduced. The central claim rests on two tuned hyperparameters, epsilon and S, and on domain-structure assumptions about disjoint single-domain clients and harmless oversampling. The equal-influence equation is an identity after resampling, not a fitted prediction.

free parameters (2)
  • Label smoothing coefficient epsilon = 0.1 on PACS per Table VI; final per-dataset values not stated
    Controls the strength of label smoothing in Eq. (1). It is chosen by sensitivity analysis rather than derived, and the paper does not report how it was selected for each benchmark.
  • Per-client training budget S = Tested values 30B, 45B, 60B with B likely the batch size; final per-dataset values not stated
    The budget determines how many samples each client trains on per round in Eq. (8). It directly drives the balancing mechanism and is tuned empirically.
assumptions (3)
  • domain assumption Each client owns exactly one domain, and the target domain is unseen and disjoint from all client domains.
    The Problem Formulation in Section III defines client i with distribution p_i and target p_T unavailable during training. The budget mechanism is motivated by equalizing clients in this specific setup and may not transfer to overlapping or multi-domain clients.
  • domain assumption Oversampling clients with fewer than S samples is harmless and preserves the benefit of balancing.
    The method resamples D_i to size S before local training. The paper provides no analysis of how repeated samples affect small-client gradients or overfitting; it relies on empirical gains in Table V.
  • standard math Standard gradient-descent and expectation-linearization identities in Eqs. (4)-(8).
    The equal-contribution argument uses linearity of expectation and treats floor(|D_i|/B) as the number of local steps. Integer and mini-batch effects are not analyzed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Federated Domain Generalization with Label Smoothing and Balanced Decentralized Training." pith.science (2026). https://pith.science/paper/4YVE7DL3

@misc{pith2026241211408,
  author       = {Pith},
  title        = {Pith review of: Federated Domain Generalization with Label Smoothing and Balanced Decentralized Training},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4YVE7DL3}},
  note         = {Machine review of arXiv:2412.11408}
}
read the original abstract

In this paper, we propose a novel approach, Federated Domain Generalization with Label Smoothing and Balanced Decentralized Training (FedSB), to address the challenges of data heterogeneity within a federated learning framework. FedSB utilizes label smoothing at the client level to prevent overfitting to domain-specific features, thereby enhancing generalization capabilities across diverse domains when aggregating local models into a global model. Additionally, FedSB incorporates a decentralized budgeting mechanism which balances training among clients, which is shown to improve the performance of the aggregated global model. Extensive experiments on four commonly used multi-domain datasets, PACS, VLCS, OfficeHome, and TerraInc, demonstrate that FedSB outperforms competing methods, achieving state-of-the-art results on three out of four datasets, indicating the effectiveness of FedSB in addressing data heterogeneity.

Figures

Figures reproduced from arXiv: 2412.11408 by the authors.

Figure 1
Figure 1. Overview of FedSB. Blue and Orange arrows show local model loading to the server and initialization with the global model from the server, respectively. privacy [1], [16]. FedAVG [1] is a very popular algorithm, where local models are iteratively uploaded to a central server, and then through a weighted averaging scheme, a global model is produced, which will then be downloaded to the clients. Many other methods bui… view at source ↗
Figure 2
Figure 2. TSNE plot for S domain on PACS. The points are color-coded to represent different classes. TABLE III: Accuracy of FedAvg and FedSB on the PACS dataset using ViT backbones. Backbone Method P A C S Ave. ViT-b/16 FedAvg [1] 98.88 90.65 76.56 82.89 87.24 FebSB (ours) 98.84 91.49 80.43 83.48 88.56 ViT-b/32 FedAvg [1] 96.66 84.57 77.46 79.01 84.47 FebSB (ours) 96.96 87.89 78.07 78.43 85.97 and train the model on the rest … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 21 canonical work pages

  1. [1]

    Communication-efficient learning of deep networks from decentralized data,

    B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, “Communication-efficient learning of deep networks from decentralized data,” in Artificial Intelligence and Statistics , pp. 1273–1282, 2017

    work page 2017

  2. [2]

    Towards unsupervised domain generalization,

    X. Zhang, L. Zhou, R. Xu, P. Cui, Z. Shen, and H. Liu, “Towards unsupervised domain generalization,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pp. 4910–4920, June 2022

    work page 2022

  3. [3]

    Learning to generate novel domains for domain generalization,

    K. Zhou, Y . Yang, T. Hospedales, and T. Xiang, “Learning to generate novel domains for domain generalization,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XVI 16 , pp. 561–578, Springer, 2020

    work page 2020

  4. [4]

    Learning from extrin- sic and intrinsic supervisions for domain generalization,

    S. Wang, L. Yu, C. Li, C.-W. Fu, and P.-A. Heng, “Learning from extrin- sic and intrinsic supervisions for domain generalization,” in European Conference on Computer Vision , pp. 159–176, Springer, 2020

    work page 2020

  5. [5]

    A fourier-based framework for domain generalization,

    Q. Xu, R. Zhang, Y . Zhang, Y . Wang, and Q. Tian, “A fourier-based framework for domain generalization,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. 14383– 14392, 2021

    work page 2021

  6. [6]

    Fedsr: A simple and effective domain generalization method for federated learning,

    A. T. Nguyen, P. Torr, and S. N. Lim, “Fedsr: A simple and effective domain generalization method for federated learning,” Advances in Neural Information Processing Systems, vol. 35, pp. 38831–38843, 2022

    work page 2022

  7. [7]

    Benchmarking Algorithms for Federated Domain Generalization

    R. Bai, S. Bagchi, and D. I. Inouye, “Benchmarking algorithms for federated domain generalization,” arXiv preprint arXiv:2307.04942 , 2023

    work page Pith review arXiv 2023

  8. [8]

    Federated domain generalization with generalization adjustment,

    R. Zhang, Q. Xu, J. Yao, Y . Zhang, Q. Tian, and Y . Wang, “Federated domain generalization with generalization adjustment,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion, pp. 3954–3963, 2023

    work page 2023

Show all 27 references
  1. [9]

    Federated learning with domain generalization,

    L. Zhang, X. Lei, Y . Shi, H. Huang, and C. Chen, “Federated learning with domain generalization,” arXiv preprint arXiv:2111.10487 , 2021

    2021 arXiv

  2. [10]

    Federated optimization in heterogeneous networks,

    T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, and V . Smith, “Federated optimization in heterogeneous networks,” Proceedings of Machine Learning and Systems , vol. 2, pp. 429–450, 2020

    2020

  3. [11]

    In search of lost domain generalization,

    I. Gulrajani and D. Lopez-Paz, “In search of lost domain generalization,” in International Conference on Learning Representations , 2021

    2021

  4. [12]

    Deeper, broader and artier domain generalization,

    D. Li, Y . Yang, Y .-Z. Song, and T. M. Hospedales, “Deeper, broader and artier domain generalization,” in Proceedings of the IEEE International Conference on Computer Vision , pp. 5542–5550, 2017

    2017

  5. [13]

    Deep hashing network for unsupervised domain adaptation,

    H. Venkateswara, J. Eusebio, S. Chakraborty, and S. Panchanathan, “Deep hashing network for unsupervised domain adaptation,” in Pro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 5018–5027, 2017

    2017

  6. [14]

    Recognition in terra incognita,

    S. Beery, G. Van Horn, and P. Perona, “Recognition in terra incognita,” in Proceedings of the European Conference on Computer Vision , pp. 456– 473, 2018

    2018

  7. [15]

    Unbiased metric learning: On the utilization of multiple datasets and web images for softening bias,

    C. Fang, Y . Xu, and D. N. Rockmore, “Unbiased metric learning: On the utilization of multiple datasets and web images for softening bias,” in Proceedings of the IEEE International Conference on Computer Vision , pp. 1657–1664, 2013

    2013

  8. [16]

    Federated learning: Strategies for improving communication efficiency,

    J. Kone ˇcn`y, H. B. McMahan, F. X. Yu, P. Richt ´arik, A. T. Suresh, and D. Bacon, “Federated learning: Strategies for improving communication efficiency,” arXiv preprint arXiv:1610.05492 , 2016

    2016 arXiv

  9. [17]

    Model-contrastive federated learning,

    Q. Li, B. He, and D. Song, “Model-contrastive federated learning,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 10713–10722, 2021

    2021

  10. [18]

    Federated dropout—a simple ap- proach for enabling federated learning on resource constrained devices,

    D. Wen, K.-J. Jeon, and K. Huang, “Federated dropout—a simple ap- proach for enabling federated learning on resource constrained devices,” IEEE wireless communications letters, vol. 11, no. 5, pp. 923–927, 2022

    2022

  11. [19]

    Tackling the ob- jective inconsistency problem in heterogeneous federated optimization,

    J. Wang, Q. Liu, H. Liang, G. Joshi, and H. V . Poor, “Tackling the ob- jective inconsistency problem in heterogeneous federated optimization,” Advances in Neural Information Processing Systems , vol. 33, pp. 7611– 7623, 2020

    2020

  12. [20]

    Federated domain generaliza- tion for image recognition via cross-client style transfer,

    J. Chen, M. Jiang, Q. Dou, and Q. Chen, “Federated domain generaliza- tion for image recognition via cross-client style transfer,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pp. 361–370, 2023

    2023

  13. [21]

    Collaborative optimization and aggregation for decentralized domain generalization and adaptation,

    G. Wu and S. Gong, “Collaborative optimization and aggregation for decentralized domain generalization and adaptation,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , pp. 6484– 6493, 2021

    2021

  14. [22]

    Fedproto: Federated prototype learning across heterogeneous clients,

    Y . Tan, G. Long, L. Liu, T. Zhou, Q. Lu, J. Jiang, and C. Zhang, “Fedproto: Federated prototype learning across heterogeneous clients,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 36, pp. 8432–8440, 2022

    2022

  15. [23]

    Contrastive- enhanced domain generalization with federated learning,

    X. Yu, D. Wang, M. J. McKeown, and Z. J. Wang, “Contrastive- enhanced domain generalization with federated learning,” IEEE Trans- actions on Artificial Intelligence , vol. 5, no. 4, pp. 1525–1532, 2023

    2023

  16. [24]

    Federated unsupervised domain generalization using global and local alignment of gradients,

    F. Pourpanah, M. Molahasani, M. Soltany, M. Greenspan, and A. Etemad, “Federated unsupervised domain generalization using global and local alignment of gradients,” arXiv preprint arXiv:2405.16304 , 2024

    2024 arXiv

  17. [25]

    Out-of-distribution generalization of federated learning via implicit invariant relationships,

    Y . Guo, K. Guo, X. Cao, T. Wu, and Y . Chang, “Out-of-distribution generalization of federated learning via implicit invariant relationships,” in International Conference on Machine Learning , pp. 11905–11933, PMLR, 2023

    2023

  18. [26]

    Feddg: Federated do- main generalization on medical image segmentation via episodic learn- ing in continuous frequency space,

    Q. Liu, C. Chen, J. Qin, Q. Dou, and P.-A. Heng, “Feddg: Federated do- main generalization on medical image segmentation via episodic learn- ing in continuous frequency space,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. 1013– 1023, 2021

    2021

  19. [27]

    An image is worth 16x16 words: Trans- formers for image recognition at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, “An image is worth 16x16 words: Trans- formers for image recognition at scale,” in International Conference on Learni...

    2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.