Pith. sign in

REVIEW 3 major objections 5 minor 37 references

FedDAA: Dynamic Client Clustering for Concept Drift Adaptation in Federated Learning

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read FedDAA distinguishes real drift from virtual/label drift in federated learning and selectively keeps historical data, reporting 7.84–8.52 percentage point accuracy gains over FedDrift.

desk verdict FedDAA is a genuinely useful combination — dynamic cluster counting plus drift-source-aware retention — but the real-drift detector in Eq. (4) rests on an invariance that the paper neither proves nor isolates in its experiments. read the letter →

arxiv 2506.21054 v1 pith:QCJDKAE3 submitted 2025-06-26 cs.LG

Fu Peng , Ming Tang This is my paper · ORCID
classification cs.LG
keywords federatedlearningconceptdriftrealdetectionvirtuallabelclustereddataprototypecatastrophicforgetting
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 federated learning system can adapt to concept drift better by first identifying which of three drift sources is occurring: real drift (a shift in $P(y|x)$), virtual drift (a shift in $P(x)$), or label drift (a shift in $P(y)$). It proposes FedDAA, which clusters clients by data prototypes and declares real drift exactly when a client's historical and current prototypes fall into different clusters. Clients flagged with real drift retrain on current data only, while clients without real drift train on both historical and current data, so useful old knowledge is not thrown away. The reported result is 7.84% to 8.52% higher average accuracy than the strongest baseline on Fashion-MNIST, CIFAR-10, and CIFAR-100, with a convergence guarantee under a dynamically changing number of clusters.

What carries the argument

The load-bearing object is the data prototype $P^t_k \in \mathbb{R}^{R\times R}$ (Definition 2), whose $r$-th column is the mean model output $h(w^t,x)$ over samples of class $r$. It compresses a client's conditional distribution $P(y|x)$ into a matrix that can be clustered. The mechanism that carries the argument is the cluster-assignment comparison of Eq. (4): previous and current prototypes are each mapped to the nearest center among the current prototype cluster centers $P^t_c$, and a change from $c^{t-1}_k$ to $c^t_k$ is the sole signal of real drift. That single test determines whether historical data are retained or discarded, so the adaptation strategy and the convergence analysis both rest on the prototype tracking conditional-distribution changes without being displaced by marginal shifts.

What would settle it

Take a two-cluster task with fixed labels and fixed $P(y|x)$, then rotate the input images in small steps (0, 5, 10, ..., 30 degrees) across time steps; if any rotation causes the client's prototype $P^t_k$ to change its nearest cluster center in Eq. (4), FedDAA will flag real drift and discard historical data, contradicting the claimed separation of virtual drift from real drift.

Watch

Extended reading notes

Core claim

The central discovery is that the drift source, not merely the fact of drift, should drive the adaptation strategy. FedDAA encodes each client's conditional distribution at time $t$ in a class-conditioned prototype matrix $P^t_k$ of averaged model outputs, clusters all clients' prototypes on the server via K-means with the cluster count chosen by silhouette scores, and labels client $k$ as having real drift if the nearest prototype-cluster center changes between $t-1$ and $t$, that is, $c^{t-1}_k \neq c^t_k$ in Eq. (4). Real-drifted clients update on current data only, whereas everyone else keeps using historical data as well, which the authors argue prevents catastrophic forgetting under virtual and label drift. The paper proves a convergence bound (Theorem 1) for the federated objective under smoothness, bounded-gradient, and bounded-dissimilarity assumptions, and the small gap to an oracle in experiments (roughly one point) is presented as evidence that the drift-source detection captures most of the benefit of full knowledge.

Load-bearing premise

The load-bearing premise is that virtual drift and label drift never move a client's data prototype across a prototype-cluster boundary, so a change in cluster assignment can be taken as proof of real drift.

Editorial extensions

If this is right

  • In deployments where clients experience virtual drift such as image rotation or label drift such as shifting class frequencies, FedDAA retains historical data and should avoid the catastrophic forgetting that single-model baselines exhibit.
  • Because the cluster count is re-estimated at every time step, the system can move from two to four decision boundaries without server-side reconfiguration of the number of clusters.
  • The accuracy gains over FedDrift (8.52% on Fashion-MNIST, 7.84% on CIFAR-10, 8.31% on CIFAR-100) imply that distinguishing real drift from virtual/label drift, rather than clustering alone, is what drives most of the improvement.
  • The sampling-rate table shows that dropping from 100% to 20% client participation costs at most about two points of accuracy, so communication cost can be cut substantially in practice.
  • The convergence bound in Theorem 1 makes the choice of communication rounds and learning rate principled: the gap to the optimal objective is bounded by a constant that shrinks with the expected number of clusters.

Reading between the lines

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

  • Editorial inference: switching the detection from a single previous time step to a short prototype history could make real-drift alarms less sensitive to transient fluctuations in the model output.
  • Editorial inference: a pure virtual-drift benchmark with rotations but no label remapping would isolate whether Eq. (4) truly avoids false real-drift alarms, since the paper's setup mixes rotation with changing label mappings.
  • Editorial inference: the Euclidean distance from a prototype to its assigned center could be used as a drift-magnitude measure, turning the binary keep-or-discard decision into a graded memory budget.
  • Editorial inference: because the prototype only needs per-class mean model outputs, the same detection rule should transfer to non-image data such as time series or text, which the paper does not test.
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 / 5 minor

Summary. The paper proposes FedDAA, a clustered federated learning framework for multi-source concept drift. It introduces three modules: a Number of Clusters Determination (NCD) module based on clustering client data prototypes, a Real Drift Local Detection (RDLD) module that flags real drift when a client's historical and current prototypes are assigned to different prototype cluster centers (Eq. (4)), and a Distribution-Aware Adaptation (DAA) module that discards historical data for real-drifted clients but retains it for virtual/label-drifted clients. The authors provide a convergence analysis (Theorem 1) and report average accuracy gains of 7.84%–8.52% over FedDrift on Fashion-MNIST, CIFAR-10, and CIFAR-100, with an ablation study and comparisons to an Oracle.

Significance. The problem addressed is relevant: distinguishing real drift from virtual and label drift in federated learning would enable principled selective retention of historical data. The paper's strengths include a clearly documented experimental protocol, ablations that isolate the contributions of the three modules, comparisons against several baselines and an Oracle, and detailed pseudocode. If the drift-detection criterion in Eq. (4) were formally justified and the convergence theorem were valid, the work would be a solid contribution to clustered FL under non-stationary distributions. However, the central detection equivalence is not established, and the convergence analysis relies on an assumption that effectively assumes the main algorithmic claim; these issues are load-bearing and prevent acceptance in the current form.

major comments (3)
  1. [§4.3, Eq. (4)] The real-drift detection criterion c^{t-1}_k != c^t_k is not justified. The prototype P^t_k in Eq. (2) is a matrix of class-conditional means of model outputs h(w^t, x), so it depends on the current model checkpoint w^t and on the input distribution through x. A pure virtual drift, such as rotating inputs, changes h(w^t, x) even if P(y|x) is fixed; both P^{t-1}_k and P^t_k are then assigned to the nearest centers among the same current cluster centers P^t_c, so a marginal shift in P(x) can move the historical and current prototypes into different clusters. The DAA module would then discard historical data, which is exactly the wrong adaptation for virtual drift. The paper does not prove that the nearest-center assignment is invariant to virtual drift or label drift, nor does it bound the probability of such a false positive.
  2. [Appendix C.1] The experimental protocol never isolates the false-positive scenario described above. At every time step, label-permutation real drift is applied simultaneously with image rotations (120 or 240 degrees) and Dirichlet label shifts, so a client whose prototype assignment flips because of rotation alone is indistinguishable from a client that truly changed P(y|x). The reported averages in Table 1 and the forgetting-mitigation results in Table 2 therefore cannot validate the claim that FedDAA distinguishes real drift from virtual/label drift. An experiment with pure virtual drift (e.g., rotations with no label permutation) is needed to test whether Eq. (4) keeps such clients in K_clean and whether historical data are retained.
  3. [§5, Assumption 5 and Appendix B.3] The convergence analysis is not established. Assumption 5 only states E[C] = C*, but the proof uses the much stronger equality E_C[L(w^0, \hat{\alpha}^0, C)] = L(w^0, \hat{\alpha}^0, C*) (Appendix B.3, Eq. (35); see also Lemma 2 in Appendix B.2). This equality does not follow from E[C] = C*; it requires L(w^0, \hat{\alpha}^0, ·) to be linear or some additional distributional assumption on C. Moreover, the NCD module selects C by silhouette scores, and no argument is given that the selected cluster count is unbiased for an optimal C*, so Assumption 5 effectively assumes the correctness of the module that the theorem is meant to validate. Finally, Theorem 1 bounds a gap between objective values at two different cluster counts; the right-hand side contains a negative term -E C* Δ^2_ξ / (4η) plus positive terms, so it is not a standard convergence bound even if the preceding steps were valid.
minor comments (5)
  1. [Appendix B.1 heading] The heading 'Approximated objective function for piratical implementation' contains a typo; 'piratical' should be 'practical'.
  2. [Appendix B.2] The word 'numebr' should be 'number' in the sentence 'for a specific cluster numebr C'.
  3. [Table 2] In the FedDrift column, the entries for the first two rows appear as '36.72 ± 0.4165.61 ± 0.2969.20 ± 0.66' and similar, with missing separators between values; the table formatting needs correction.
  4. [Algorithm 3 title] The algorithm title 'FL Distribution-Awared Adaptation' contains a typo; 'Awared' should be 'Aware'.
  5. [§4.3] The description of the RDLD module refers to the 'the the prototype cluster centers' in the first sentence; the duplicated article should be removed.

Circularity Check

1 steps flagged · score 3.0 of 10

The empirical accuracy claims are externally evaluated and not circular, but the convergence analysis is partially circular: Assumption 5 simply assumes the NCD module's cluster count is unbiased (EC=C*), and the proof uses exactly that assumption to justify the theorem's bound.

  1. other [Section 5, Assumption 5; proof of Lemma 2 (Eq. 22) and Theorem 1 (Eq. 35)]
    "Assumption 5 (Expectation of the Number of Clusters). There is an optimal number of clusters C∗. Assume that the number of clusters C follows a certain distribution, and EC=C∗. ... Under Assumption 5, we obtain EC[L(w0, α0, C)] = L(w0, α0, C∗). ... Theorem 1 shows that using the number of clusters C determined by FedDAA, ECEξ[L(wτ , ˆατ , C∗)− L(wτ , ˆατ , C)] can be bounded by a constant, indicating the convergence of FedDAA."

    The convergence theorem is presented as a guarantee for FedDAA's dynamic cluster-number determination, but its proof assumes the very property that the NCD module is supposed to deliver: that the random cluster count C is unbiased for the true count C*. The transition EC[L(w0, α0, C)] = L(w0, α0, C*) is valid only under Assumption 5, and no argument from the NCD's silhouette-score procedure establishes this unbiasedness. Thus the claimed bound ECEξ[L(wτ, ατ, C*) − L(wτ, ατ, C)] ≤ ... reduces to an assumed version of the theorem's own target, making the convergence analysis circular with respect to RQ1.

full rationale

The central empirical claim is not circular: Tables 1-5 report held-out test accuracy against external baselines (FedAvg, A-FedAvg, Flash, FedRC, FedDrift, Oracle), and no fitted parameter is relabeled as a prediction. The main circular element is confined to the convergence analysis: Assumption 5 assumes the number-of-clusters estimate is unbiased, and Lemma 2 and Theorem 1 both use EC[L(w0, α0, C)] = L(w0, α0, C*) to convert the expectation over C into an evaluation at C*. This grants rather than proves the correctness of the NCD module. The Eq. (4) real-drift detector (cluster-assignment change) is an unsupported operational identification, not a circular one: Definition 1 defines real drift as a shift in P(y|x), while the prototype in Eq. (2) is a class-conditional mean of model outputs h(w^t, x), so virtual drift can alter the prototype and flip cluster membership without any change in P(y|x). That is a correctness risk under rotation-based virtual drift, but the paper does not define real drift as cluster-assignment change, and its empirical evaluation is external to the detector's internal definition. Therefore the appropriate circularity score is modest, reflecting the circular assumption in the theory rather than the independent empirical results.

Assumptions & free parameters 3 free parameters · 4 assumptions · 1 invented entities

The central detection mechanism assumes an unproved equivalence between prototype cluster assignment changes and real drift, and the convergence theorem assumes the NCD module selects an unbiased cluster count. The data prototype itself is an invented representation with no external falsifiable handle. These are the main uncharged premises.

free parameters (3)
  • Maximum cluster count M = not reported
    Algorithm 1 requires a maximal number of clusters M but the paper does not state its value; the chosen cluster count C^t is searched from 2 to M, so this affects NCD and drift detection.
  • Learning rate for FedDAA = 0.06
    Tuned from {0.01, 0.03, 0.06, 0.1}; a standard hyperparameter needed to reproduce Table 1 results.
  • Client sampling rate = 0.5
    Used in the main experiments; its effect is tested in Table 4, but the value is a hand-set choice.
assumptions (4)
  • standard math Assumptions 1-4: L-smoothness, bounded gradients, unbiased gradients with bounded variance, and bounded dissimilarity of local objectives.
    These are standard non-convex FL analysis assumptions, used in Theorem 1 and the appendix.
  • ad hoc to paper Assumption 5: there is an optimal number of clusters C*, and the cluster count C follows a distribution with E[C] = C*.
    This assumes the NCD module selects an unbiased cluster count in expectation, but no property of silhouette selection is used to establish it in the paper.
  • domain assumption The data prototype P^t_k in Definition 2 captures the conditional distribution P(y|x) of a client's local dataset.
    The prototype is a class-conditional mean of model outputs h(w^t, x); no formal argument shows it is a sufficient statistic for P(y|x), and it depends on the model being trained.
  • domain assumption A change in prototype cluster assignment implies real drift, and no change implies virtual or label drift.
    This is the core detection rule in Section 4.3, Eq. (4) and Algorithm 2; the paper does not prove that virtual or label drift leaves the nearest cluster center unchanged.
invented entities (1)
  • Data prototype P^t_k
    purpose: A compact matrix of class-conditional average model outputs, used for clustering clients and detecting real drift.
    It is an internal representation computed from model outputs with no external falsifiable prediction; its validity rests on the unproved equivalence between prototype shifts and drift types.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FedDAA: Dynamic Client Clustering for Concept Drift Adaptation in Federated Learning." pith.science (2026). https://pith.science/paper/QCJDKAE3

@misc{pith2026250621054,
  author       = {Pith},
  title        = {Pith review of: FedDAA: Dynamic Client Clustering for Concept Drift Adaptation in Federated Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QCJDKAE3}},
  note         = {Machine review of arXiv:2506.21054}
}
read the original abstract

In federated learning (FL), the data distribution of each client may change over time, introducing both temporal and spatial data heterogeneity, known as concept drift. Data heterogeneity arises from three drift sources: real drift (a shift in the conditional distribution P(y|x)), virtual drift (a shift in the input distribution P(x)), and label drift (a shift in the label distribution P(y)). However, most existing FL methods addressing concept drift primarily focus on real drift. When clients experience virtual or label drift, these methods often fail to selectively retain useful historical knowledge, leading to catastrophic forgetting. A key challenge lies in distinguishing different sources of drift, as they require distinct adaptation strategies: real drift calls for discarding outdated data, while virtual or label drift benefits from retaining historical data. Without explicitly identifying the drift sources, a general adaptation strategy is suboptimal and may harm generalization. To address this challenge, we propose FedDAA, a dynamic clustered FL framework designed to adapt to multi-source concept drift while preserving valuable historical knowledge. Specifically, FedDAA integrates three modules: a cluster number determination module to find the optimal number of clusters; a real drift detection module to distinguish real drift from virtual/label drift; and a concept drift adaptation module to adapt to new data while retaining useful historical information. We provide theoretical convergence guarantees, and experiments show that FedDAA achieves 7.84% to 8.52% accuracy improvements over state-of-the-art methods on Fashion-MNIST, CIFAR-10, and CIFAR-100.

Figures

Figures reproduced from arXiv: 2506.21054 by the authors.

Figure 1
Figure 1. FL under con￾cept drift. There are two di￾mensional data heterogeneity: time and space. “A”, “B” and “C” represent different data dis￾tributions. To deal with data heterogeneity in space (i.e., across clients), existing works [4–13] proposed clustered FL methods. Clustered FL is a framework that groups clients into clusters based on their local data distributions to address the data heterogeneity across clients. The… view at source ↗
Figure 2
Figure 2. An overview of the proposed framwork FedDAA. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Visualization of the concept drift settings. [PITH_FULL_IMAGE:figures/full_fig_p021_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Average training accuracy of all methods on CIFAR-10 over 6 time steps. Each time step [PITH_FULL_IMAGE:figures/full_fig_p022_4.png]
Figure 5
Figure 5. Figure 5: Average test accuracy on CIFAR-10 for all tasks over 6 time steps. Each time step contains [PITH_FULL_IMAGE:figures/full_fig_p022_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 33 canonical work pages

  1. [1]

    Learning under concept drift: A review.IEEE Trans

    Jie Lu, Anjin Liu, Fan Dong, Feng Gu, João Gama, and Guangquan Zhang. Learning under concept drift: A review.IEEE Trans. Knowl. Data Eng., 31(12):2346–2363, 2019

    work page 2019

  2. [2]

    Brendan McMahan, Brendan Avent, et al

    Peter Kairouz, H. Brendan McMahan, Brendan Avent, et al. Advances and open problems in federated learning.Found. Trends Mach. Learn., 14(1-2):1–210, 2021

    work page 2021

  3. [3]

    A survey on concept drift adaptation.ACM Comput

    João Gama, Indre Zliobaite, Albert Bifet, Mykola Pechenizkiy, and Abdelhamid Bouchachia. A survey on concept drift adaptation.ACM Comput. Surv., 46(4):44:1–44:37, 2014

    work page 2014

  4. [4]

    Fedrc: Tackling diverse distribution shifts challenge in federated learning by robust clustering

    Yongxin Guo, Xiaoying Tang, and Tao Lin. Fedrc: Tackling diverse distribution shifts challenge in federated learning by robust clustering. InProceedings of Forty-first International Conference on Machine Learning, ICML 2024, 2024

    work page 2024

  5. [5]

    An efficient framework for clustered federated learning

    Avishek Ghosh, Jichan Chung, Dong Yin, and Kannan Ramchandran. An efficient framework for clustered federated learning. InProceedings of Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, 2020

    work page 2020

  6. [6]

    Accelerat- ing federated learning with cluster construction and hierarchical aggregation.IEEE Trans

    Zhiyuan Wang, Hongli Xu, Jianchun Liu, Yang Xu, He Huang, and Yangming Zhao. Accelerat- ing federated learning with cluster construction and hierarchical aggregation.IEEE Trans. Mob. Comput., 22(7):3805–3822, 2023

    work page 2023

  7. [7]

    Federated learning with hierarchical cluster- ing of local updates to improve training on non-iid data

    Christopher Briggs, Zhong Fan, and Peter Andras. Federated learning with hierarchical cluster- ing of local updates to improve training on non-iid data. InProceedings of 2020 International Joint Conference on Neural Networks, IJCNN 2020, pages 1–9, 2020

    work page 2020

  8. [8]

    Clustered federated learning: Model- agnostic distributed multitask optimization under privacy constraints.IEEE Trans

    Felix Sattler, Klaus-Robert Müller, and Wojciech Samek. Clustered federated learning: Model- agnostic distributed multitask optimization under privacy constraints.IEEE Trans. Neural Networks Learn. Syst., 32(8):3710–3722, 2021

    work page 2021

Show all 37 references
  1. [9]

    Fedsoft: Soft clustered federated learning with proximal local updating

    Yichen Ruan and Carlee Joe-Wong. Fedsoft: Soft clustered federated learning with proximal local updating. InProceedings of Thirty-Sixth AAAI Conference on Artificial Intelligence, AAAI 2022, Thirty-Fourth Conference on Innovative Applications of Artificial Intelligence, IAAI 2...

    2022

  2. [10]

    Federated multi-task learning under a mixture of distributions

    Othmane Marfoq, Giovanni Neglia, Aurélien Bellet, Laetitia Kameni, and Richard Vidal. Federated multi-task learning under a mixture of distributions. InProceedings of Advances in Neural Information Processing Systems 34: Annual Conference on Neural Information Processing Syste...

    2021

  3. [11]

    Multi-center federated learning: clients clustering for better personalization.World Wide Web (WWW), 26(1): 481–500, 2023

    Guodong Long, Ming Xie, Tao Shen, Tianyi Zhou, Xianzhi Wang, and Jing Jiang. Multi-center federated learning: clients clustering for better personalization.World Wide Web (WWW), 26(1): 481–500, 2023

    2023

  4. [12]

    Towards federated clustering: A federated fuzzy c-means algorithm (FFCM).CoRR, abs/2201.07316, 2022

    Morris Stallmann and Anna Wilbik. Towards federated clustering: A federated fuzzy c-means algorithm (FFCM).CoRR, abs/2201.07316, 2022

    2022 arXiv

  5. [13]

    Clustered federated learning in heterogeneous environment.IEEE Trans

    Yihan Yan, Xiaojun Tong, and Shen Wang. Clustered federated learning in heterogeneous environment.IEEE Trans. Neural Networks Learn. Syst., 35(9):12796–12809, 2024

    2024

  6. [14]

    Ellango Jothimurugesan, Kevin Hsieh, Jianyu Wang, Gauri Joshi, and Phillip B. Gibbons. Federated learning under distributed concept drift. InProceedings of International Conference on Artificial Intelligence and Statistics, volume 206 ofProceedings of Machine Learning Research...

    2023

  7. [15]

    A multi-model approach for handling concept drifting data in federated learning

    Guanhui Yang, Xiaoting Chen, Tengsen Zhang, and Shuo Wang. A multi-model approach for handling concept drifting data in federated learning. InWorkshop of Distributed Machine Learning and Unlearning for Sensor-Cloud Systems (DLS2) in the 20th International Conference on Mobilit...

    2024

  8. [16]

    Classifier clustering and feature alignment for federated learning under distributed concept drift

    Junbao Chen, Jingfeng Xue, Yong Wang, Zhenyan Liu, and Lu Huang. Classifier clustering and feature alignment for federated learning under distributed concept drift. InProceedings of Advances in Neural Information Processing Systems 38: Annual Conference on Neural Information P...

    2024

  9. [17]

    Drift detection and adaptation for federated learning in iot with adaptive device management

    Shuang Zhou, Shashank Shekhar, Ajay Chhokra, Abhishek Dubey, and Aniruddha Gokhale. Drift detection and adaptation for federated learning in iot with adaptive device management. In2024 IEEE International Conference on Big Data (BigData), pages 8088–8097. IEEE, 2024

    2024

  10. [18]

    Casado, Dylan Lema, Marcos F

    Fernando E. Casado, Dylan Lema, Marcos F. Criado, Roberto Iglesias, Carlos Vázquez Regueiro, and Senén Barro. Concept drift detection and adaptation for federated and continual learning. Multim. Tools Appl., 81(3):3397–3419, 2022

    2022

  11. [19]

    Asynchronous federated learning for sensor data with concept drift

    Yujing Chen, Zheng Chai, Yue Cheng, and Huzefa Rangwala. Asynchronous federated learning for sensor data with concept drift. InProceedings of 2021 IEEE International Conference on Big Data (Big Data), pages 4822–4831, 2021

    2021

  12. [20]

    Federating from history in streaming federated learning

    Ruirui Zhang, Yifei Zou, Zhenzhen Xie, Xiao Zhang, Peng Li, Zhipeng Cai, Xiuzhen Cheng, and Dongxiao Yu. Federating from history in streaming federated learning. InProceedings of the Twenty-fifth International Symposium on Theory, Algorithmic Foundations, and Protocol Design f...

    2024

  13. [21]

    Adaptive federated learning in presence of concept drift

    Giuseppe Canonaco, Alex Bergamasco, Alessio Mongelluzzo, and Manuel Roveri. Adaptive federated learning in presence of concept drift. InProceedings of International Joint Conference on Neural Networks, IJCNN 2021, pages 1–7, 2021

    2021

  14. [22]

    Flash: Concept drift adaptation in federated learning

    Kunjal Panchal, Sunav Choudhary, Subrata Mitra, Koyel Mukherjee, Somdeb Sarkhel, Saayan Mitra, and Hui Guan. Flash: Concept drift adaptation in federated learning. InInternational Conference on Machine Learning, ICML 2023, volume 202 ofProceedings of Machine Learning Research,...

    2023

  15. [23]

    Client-side adaptation to concept drift in federated learning

    Finn Saile, Julius Thomas, Dominik Kaaser, and Stefan Schulte. Client-side adaptation to concept drift in federated learning. InProceedings of 2nd International Conference on Federated Learning Technologies and Applications (FLTA), pages 71–78. IEEE, 2024

    2024

  16. [24]

    On the convergence of A class of adam-type algorithms for non-convex optimization

    Xiangyi Chen, Sijia Liu, Ruoyu Sun, and Mingyi Hong. On the convergence of A class of adam-type algorithms for non-convex optimization. InProceedings of The 7th International Conference on Learning Representations, ICLR 2019, 2019

    2019

  17. [25]

    On the almost sure convergence of stochastic gradient descent in non-convex problems

    Panayotis Mertikopoulos, Nadav Hallak, Ali Kavis, and V olkan Cevher. On the almost sure convergence of stochastic gradient descent in non-convex problems. InProceedings of Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing...

    2020

  18. [26]

    On the convergence of fedavg on non-iid data

    Xiang Li, Kaixuan Huang, Wenhao Yang, Shusen Wang, and Zhihua Zhang. On the convergence of fedavg on non-iid data. InProceedings of The 8th International Conference on Learning Representations, ICLR 2020, 2020

    2020

  19. [27]

    Vincent Poor

    Jianyu Wang, Qinghua Liu, Hao Liang, Gauri Joshi, and H. Vincent Poor. Tackling the objective inconsistency problem in heterogeneous federated optimization. InProceedings of Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processin...

    2020

  20. [28]

    Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms.CoRR, abs/1708.07747, 2017

    Han Xiao, Kashif Rasul, and Roland V ollgraf. Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms.CoRR, abs/1708.07747, 2017

    2017 arXiv

  21. [29]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny images. 2009. 11

    2009

  22. [30]

    Deep residual learning for im- age recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for im- age recognition. InProceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, pages 770–778, 2016

    2016

  23. [31]

    Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, and Hartwig Adam

    Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, and Hartwig Adam. Mobilenets: Efficient convolutional neural networks for mobile vision applications.CoRR, abs/1704.04861, 2017

    2017 arXiv

  24. [32]

    Communication-efficient learning of deep networks from decentralized data

    Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, and Blaise Agüera y Arcas. Communication-efficient learning of deep networks from decentralized data. InProceedings of the 20th International Conference on Artificial Intelligence and Statistics, AISTATS 2017, volume 5...

    2017

  25. [33]

    Tuan Nguyen, Toan Tran, Yarin Gal, Philip H

    A. Tuan Nguyen, Toan Tran, Yarin Gal, Philip H. S. Torr, and Atilim Gunes Baydin. KL guided domain adaptation. InProceedings of The Tenth International Conference on Learning Representations, ICLR 2022, 2022

    2022

  26. [34]

    Tuan Nguyen, Philip H

    A. Tuan Nguyen, Philip H. S. Torr, and Ser Nam Lim. Fedsr: A simple and effective domain generalization method for federated learning. InProceedings of Advances in Neural Information Processing Systems 35: Annual Conference on Neural Information Processing Systems 2022, NeurIP...

    2022

  27. [35]

    Information-theoretic analysis of unsupervised domain adapta- tion

    Ziqiao Wang and Yongyi Mao. Information-theoretic analysis of unsupervised domain adapta- tion. InProceedings of The Eleventh International Conference on Learning Representations, ICLR 2023, 2023

    2023

  28. [36]

    Non-stationary domain generalization: Theory and algorithm

    Thai-Hoang Pham, Xueru Zhang, and Ping Zhang. Non-stationary domain generalization: Theory and algorithm. InUncertainty in Artificial Intelligence, volume 244, pages 2902–2927, 2024

    2024

  29. [37]

    E 4η CX c=1 Eξ w∗ c −w 0 c 2 +E ξ L(w0,ˆα0, C) # (34) =L(w ∗,ˆα∗, C∗)− L(w0,ˆα0, C∗) + 2(Eη2L(ηL+ 1)σ 2 + 2E3η3L2G2) − E 4η EC

    Tzu-Ming Harry Hsu, Hang Qi, and Matthew Brown. Measuring the effects of non-identical data distribution for federated visual classification.CoRR, abs/1909.06335, 2019. 12 A Algorithm A.1 Number of Clusters Determination Algorithm 1 presents the pseudocode of the NCD module at...

    1909 arXiv

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.