Pith. sign in

REVIEW 3 major objections 6 minor 40 references

Encryption-Compatible Clustered Federated Learning via Distributed Expectation-Maximization over Metadata

T0 review · 3 major / 6 minor · reviewed 2026-07-31 · grok-4.5

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

Pith's one-line read Metadata-based clustered federated learning can stay private if the server is limited to additive updates via distributed EM.

desk verdict Solid systems fix: distributed EM makes metadata CFL additive-only and competitive in plaintext; the joint encryption-plus-accuracy claim still needs end-to-end numbers. read the letter →

arxiv 2607.28338 v1 pith:HJEXZDTH submitted 2026-07-30 cs.LG cs.CRcs.DCstat.ML

Michael Ben Ali , Imen Megdiche , André Péninou , Olivier Teste This is my paper
classification cs.LGcs.CRcs.DCstat.ML
keywords clusteredfederatedlearningexpectation-maximizationmetadatasecureaggregationhomomorphicencryptionnon-IIDdataCFLtrilemmarandomneuralprojections
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

Clustered federated learning groups clients with similar data so each group can train a better specialized model, but existing designs force a three-way trade-off among privacy, communication, and computation—the CFL trilemma. Metadata methods are cheap to run and communicate, yet they usually need plaintext dataset summaries and clash with standard secure aggregation and homomorphic encryption. This paper argues that the conflict is not inevitable: if clustering is cast as a distributed Expectation-Maximization procedure over low-dimensional metadata, every non-linear step stays on the client and the server only adds vectors. The resulting method, FLAMECHE, uses fixed random-network class-wise means as metadata, hard cosine assignments, and additive centroid updates, so it plugs into ordinary secure FL schemes without changing those protocols. Across several image datasets and hard heterogeneity mixes, it matches or beats prior CFL families in accuracy while keeping clustering cost linear in clients and metadata size rather than quadratic in model size.

What carries the argument

FLAMECHE: a distributed EM loop in which clients compute cluster responsibilities (or hard cosine assignments) from static low-dimensional metadata and broadcast parameters, then the server (or clients after additive aggregation) updates mixture parameters from summed sufficient statistics only—exactly the operations Secure Aggregation, Paillier, and CKKS handle efficiently.

What would settle it

On the paper’s own combined concept-shift-plus-label-skew suites, replace the random extractor with noise that destroys angular structure (or force soft/GMM responsibilities under heavy CKKS noise) and check whether ARI collapses and FLAMECHE accuracy falls below strong server-side or IFCA baselines while encrypted additive aggregation still runs.

Watch

Extended reading notes

Core claim

Reformulating metadata-based clustered federated learning as distributed EM over exponential-family (or spherical k-means) sufficient statistics confines the server to additive aggregation of optionally encrypted weighted metadata. That single design choice makes efficient metadata clustering compatible with practical secure FL mechanisms and, with zero-shot random latent class means, yields stronger and more stable client models under diverse non-IID shifts than server-side, client-side, or prior metadata baselines.

Load-bearing premise

The method needs same-distribution samples to sit at smaller angles than different-distribution ones, and a randomly initialized ReLU network to preserve those angles well enough that class-wise means separate the true client groups.

Editorial extensions

If this is right

  • Metadata-based CFL no longer has to choose between efficiency and standard additive encryption; both can be kept at once.
  • Clustering quality and K can be scored on static encrypted metadata (e.g., DBI) before any federated training rounds.
  • Server clustering cost scales with metadata dimension and K, not with full model size or pairwise client similarities.
  • Hard cosine EM over random class-wise means is enough to approach oracle cluster accuracy on the tested image heterogeneities.

Reading between the lines

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

  • The same additive-sufficient-statistic pattern could extend to soft GMM or other exponential-family metadata once partial-participation EM convergence is settled.
  • If random angular preservation fails on non-vision modalities, the EM shell can keep encryption compatibility while swapping in a different fixed extractor.
  • Persistent cross-round metadata under Secure Aggregation may push deployments toward per-round participant-only M-steps, a systems constraint separate from the learning claim.
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 paper identifies a CFL trilemma (privacy vs. computation vs. communication) and argues that metadata-based clustered FL is efficient but incompatible with standard secure FL schemes because clustering uses non-additive server operations. It proposes FLAMECHE: metadata-based client clustering reformulated as distributed EM over exponential-family (practically hard spherical k-means with cosine) sufficient statistics, so the server only performs additive aggregations while clients run the non-linear E-step. Metadata are class-wise means in a shared randomly initialized feature extractor (zero-shot). The method is claimed compatible with Secure Aggregation, Paillier, and CKKS without changing those protocols. Experiments on five datasets under three heterogeneity regimes compare against server-side, client-side, and metadata baselines; FLAMECHE matches or approaches an oracle partition and ranks best overall on accuracy, with ablations on metadata design, missing labels, reclustering, partial vs. accumulated M-step, and CKKS noise.

Significance. If the joint claim holds, the work closes a real gap: metadata CFL has been efficient but privacy-weak, and encryption-compatible clustering has been considered expensive. Restricting the server to additive M-steps over low-dimensional metadata is a clean systems insight that fits existing SecAgg/HE pipelines and is supported by complexity arguments (Table 1) and crypto microbenchmarks (App. D). The multi-dataset, multi-shift evaluation with seeds, aligned ranks, and several ablations is stronger than typical CFL papers. The trilemma framing is useful even if partly rhetorical. The main contribution is therefore empirical/systems rather than theoretical (convergence under partial participation is left open).

major comments (3)
  1. [§5 Table 2; App. C.3, D.1, Tables 9–10] The central joint claim (encryption-compatible metadata CFL that still improves client accuracy / trilemma position) is only partially validated. Table 2 and Fig. 3 report accuracy under plaintext aggregation with M-step over the full seen-client set V and exact K=4 (Alg. 1, §5). Appendix C.3 and D.1 state that standard Secure Aggregation forces M-step on current-round participants only; Table 9 gives only ARI/DBI for that variant, not final test accuracy. Table 10 shows CKKS noise already drops TissueMNIST ARI from 1.00 to 0.85±0.17, again without end-to-end accuracy. Without accuracy under the encryption-compatible operating points (partial M-step and/or CKKS/Paillier), the abstract/title claim reduces to “compatible in design, accurate in plaintext.” Please add end-to-end accuracy tables for those regimes, or explicitly qualify the joint claim and trilemma positioning.
  2. [§5; Appendix A–B] K is fixed to the ground-truth number of latent groups (K=4) for all methods that need it (§A, “Choice of the Number of Clusters”). Appendix B shows DBI can guide K from metadata, but main results do not use selected K under misspecification. Because FLAMECHE’s advantage is partly early structure discovery in metadata space, sensitivity to over/under-specified K (and comparison when baselines also use DBI/threshold search without oracle K) is load-bearing for the claim of robust superiority over server-side methods that also struggle under combined shifts. A controlled K-misspecification experiment on at least one dataset would strengthen §5.
  3. [§3.3 Assumption 2; Table 3] Assumption 2 (§3.3) and the random-ReLU class-wise-mean metadata are the linchpin of zero-shot clustering. Table 3 supports strong ARI for Gaussian/Uniform projections, but the assumption is stated as a condition rather than checked (e.g., angular separation statistics before/after F). Given that Mean (raw class-wise average images) is competitive on several datasets while PCA/PACFL collapse under additive EM, the paper should clarify when random latent means are necessary versus when simpler plaintext-sensitive statistics suffice, and whether Assumption 2 fails on any evaluated shift. This is needed to bound the scope of the zero-shot design claim.
minor comments (6)
  1. [§3.1] §3.1: “differs from CFM” appears to be a typo for CFL.
  2. [Figure 1] Figure 1 is described but the trilemma edges are not quantitatively mapped to methods beyond qualitative placement; a short caption tying each family to the sacrificed axis would help.
  3. [Table 1] Table 1 footnote says server-side/metadata-based omit algorithm-dependent clustering cost, which makes the comparison slightly asymmetric; note this explicitly in the table caption.
  4. [Algorithm 1; §4] Algorithm 1 mixes model training and clustering; a one-line note that clustering is independent of w_k (as claimed in §4) would reduce confusion when reading the pseudocode.
  5. [§2.2] Related work on FedEM [11] and FeSEM [22] is clear; a brief sentence on how FLAMECHE differs from other encrypted clustering literature beyond Zhang et al. [38] would round out §2.2.
  6. [Table 2; §5] PathMNIST in Table 2: FLAMECHE is slightly below K-Fed and FedGroup on pooled accuracy; the text says “highest overall performance in 4 out of 5 datasets,” which is fine, but call out PathMNIST explicitly to avoid over-reading the abstract.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: algorithmic design plus external empirical benchmarks, not a self-referential derivation.

full rationale

FLAMECHE is proposed as an engineering reformulation (metadata-based CFL cast as distributed EM so the server only aggregates additive sufficient statistics), then evaluated against external baselines (FedAvg, FedGroup, StoCFL, FeSEM, IFCA, K-Fed, PACFL) and an Oracle partition on held-out accuracy, ARI, and DBI. The EM restriction to exponential-family mixtures is an explicit design choice that makes the M-step additive by construction of the algorithm, not a claimed first-principles prediction that collapses onto fitted inputs. Assumption 2 is justified by an external citation (Giryes et al., 2016) and checked empirically in Table 3; K is fixed to the known experimental construction or guided by unsupervised DBI on metadata as a heuristic, not as a circular forecast of the main accuracy claim. The authors’ own survey is cited only for CFL taxonomy, not as a uniqueness theorem forcing the result. No step equates a fitted quantity to a ‘prediction’ of itself, and no load-bearing uniqueness or ansatz is imported from overlapping prior work. Standard systems/ML evaluation against external methods; circularity score 0.

Assumptions & free parameters 5 free parameters · 5 assumptions · 3 invented entities

The central claim rests on standard EM/exponential-family aggregation facts, FL secure-aggregation/HE additive constraints, a geometric assumption about random ReLU features, and several experimental design choices (known K, hard cosine assignments, fixed random extractor). No new physical entities; the main invented constructs are the trilemma framing and the FLAMECHE procedure itself.

free parameters (5)
  • Number of clusters K = 4
    Fixed to 4 to match constructed latent groups for all methods that need K; selection is not learned from encrypted data in the main results.
  • Reclustering period τ = 10
    Hand-chosen frequency for empty-cluster repair; ablated at 10/20/30 but main runs use a fixed value.
  • Client participation rate p = 0.2
    System parameter fixed for all comparisons; affects EM stability under partial participation.
  • Random feature-extractor initialization = Kaiming Gaussian shared seed
    Shared Kaiming Gaussian (or uniform in ablation) seed defines the metadata space; representation quality depends on this draw though reported as stable across seeds.
  • Local training hyperparameters = 5/10 epochs, lr=1e-3, 100 rounds
    Epochs, Adam lr, batch size, rounds fixed identically across methods; they affect absolute accuracy though not the clustering mechanism itself.
assumptions (5)
  • standard math Mixture components in the exponential family have M-steps that reduce to additive aggregation of responsibility-weighted sufficient statistics.
    Invoked as Assumption 1 and classical EM facts (§3.2; Dempster et al., Murphy).
  • domain assumption Practical secure FL schemes (secure aggregation, Paillier, CKKS) efficiently support additive/linear server operations and become costly or unsupported for general non-linear clustering.
    Used throughout §2.2, §3 intro, §4, and Appendix D to justify additive-only server design.
  • domain assumption Same-distribution samples have smaller angular separation than different-distribution samples, and random ReLU nets approximately preserve angles so class-wise latent means separate client groups.
    Assumption 2 (§3.3), justified by citation to Giryes et al. rather than proved for the paper’s datasets.
  • ad hoc to paper Hard spherical k-means-style assignment with cosine dissimilarity is an adequate practical limit case of GMM EM for this metadata.
    Chosen in §3.4 for efficiency and encryption simplicity; soft GMM left to future work.
  • domain assumption Clients can be modeled as arising from K latent distributions with K known and K < N.
    Standard CFL modeling assumption restated in §3 and fixed in experiments (Appendix A).
invented entities (3)
  • CFL trilemma (privacy vs computation vs communication)
    purpose: Organize prior CFL families and claim FLAMECHE improves the metadata edge by restoring encryption compatibility.
    Conceptual framing introduced in §1/Figure 1; useful taxonomy rather than a falsifiable physical entity.
  • FLAMECHE algorithm (distributed EM over hidden metadata with additive server M-step)
    purpose: Concrete protocol realizing encryption-compatible metadata-based CFL under partial participation.
    Primary methodological contribution (§3–4, Algorithm 1); evidence is empirical accuracy/ARI and crypto microbenchmarks in this paper.
  • Zero-shot class-wise mean metadata from shared random feature extractor
    purpose: Provide stable low-dimensional client signatures without pretraining or full-participation alignment rounds.
    Design choice in §3.3; supported by ablation Table 3 but not independently standardized outside this work.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Encryption-Compatible Clustered Federated Learning via Distributed Expectation-Maximization over Metadata." pith.science (2026). https://pith.science/paper/HJEXZDTH

@misc{pith2026260728338,
  author       = {Pith},
  title        = {Pith review of: Encryption-Compatible Clustered Federated Learning via Distributed Expectation-Maximization over Metadata},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HJEXZDTH}},
  note         = {Machine review of arXiv:2607.28338}
}
read the original abstract

Clustered Federated Learning (CFL) addresses data heterogeneity in federated settings by grouping clients with similar data distributions to enable effective training. Existing methods face a trade-off between privacy preservation, communication cost, and computational efficiency. We formalize this as the CFL trilemma, according to which improving two of these dimensions comes at the expense of the third. A prominent paradigm relies on metadata (i.e., low-dimensional representations of client datasets shared with the server) to enable communication- and computation-efficient clustering. However, such approaches are not compatible with standard FL privacy-preserving mechanisms. To address this limitation, we propose FLAMECHE, which reformulates metadata-based CFL as a distributed Expectation-Maximization (EM) procedure, restricting server updates to additive operations while preserving efficiency. This design enables compatibility with practical secure FL schemes. We conducted extensive experiments on multiple datasets under various heterogeneous scenarios. Results show that FLAMECHE improves the effectiveness of client models. It enables encryption-compatible metadata-based clustering, enhancing its positioning within the CFL trilemma.

Figures

Figures reproduced from arXiv: 2607.28338 by the authors.

Figure 1
Figure 1. The CFL Trilemma is gov￾erned by three competing constraints: Privacy, Computation, and Communi￾cation. As clustering relies on non-linear operations that are costly under encryption, metadata-based ap￾proaches conflict with secure FL mechanisms. These are most efficient under additive-only compu￾tations [5, 7, 38], while supporting complex operations inflate computational cost, highlighting the tension of the trile… view at source ↗
Figure 2
Figure 2. Algorithm 1 clustering mechanism. During initialization, the server initializes centroids [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Aligned ranking [21] across all datasets and heterogeneity settings. FLAMECHE achieves the best overall rank, maintaining consistent performance across concept shift on features (CSF) with label skew, concept shift on labels (CSL), and features distribution skew (FDS). Baselines that perform well on isolated shifts (e.g., PACFL) suffer severe degradation under combined heterogeneity. 6 Ablation Study: Impact of Meta… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Hyperparameter selection before FL training. The Davies-Bouldin Index (DBI) is evaluated [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]
Figure 5
Figure 5. Figure 5: Accuracy per round under concept shift on features with missing labels under CIFAR10 [PITH_FULL_IMAGE:figures/full_fig_p016_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

40 extracted references · 1 linked inside Pith

  1. [1]

    Deep learning with differential privacy

    Martin Abadi, Andy Chu, Ian Goodfellow, H Brendan McMahan, Ilya Mironov, Kunal Talwar, and Li Zhang. Deep learning with differential privacy. In Proceedings of the 2016 ACM SIGSAC conference on computer and communications security, pages 308–318, 2016

    2016

  2. [2]

    k-means++: The advantages of careful seeding

    David Arthur, Sergei Vassilvitskii, et al. k-means++: The advantages of careful seeding. In Soda, volume 7, pages 1027–1035, 2007

    2007

  3. [3]

    A systematic survey on clustering in federated learning

    Zouheir Belfeki, Moez Krichen, and Salah Zidi. A systematic survey on clustering in federated learning. Multimedia Tools and Applications, 85(5):429, 2026

    2026

  4. [4]

    A survey on clustered federated learning: Taxonomy, analysis and applications

    Michael Ben Ali, Omar El-Rifai, Imen Megdiche, André Peninou, and Olivier Teste. A survey on clustered federated learning: Taxonomy, analysis and applications. arXiv preprint arXiv:2501.17512, 2025

    arXiv 2025

  5. [5]

    Practical secure aggregation for privacy-preserving machine learning

    Keith Bonawitz, Vladimir Ivanov, Ben Kreuter, Antonio Marcedone, H Brendan McMahan, Sarvar Patel, Daniel Ramage, Aaron Segal, and Karn Seth. Practical secure aggregation for privacy-preserving machine learning. In proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security, pages 1175–1191, 2017

    2017

  6. [6]

    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. In 2020 International Joint Conference on Neural Networks (IJCNN), pages 1–9. IEEE, 2020

    2020

  7. [7]

    Homomorphic encryption for arithmetic of approximate numbers

    Jung Hee Cheon, Andrey Kim, Miran Kim, and Yongsoo Song. Homomorphic encryption for arithmetic of approximate numbers. In International conference on the theory and application of cryptology and information security, pages 409–437. Springer, 2017

    2017

  8. [8]

    Maximum likelihood from incomplete data via the em algorithm

    Arthur P Dempster, Nan M Laird, and Donald B Rubin. Maximum likelihood from incomplete data via the em algorithm. Journal of the royal statistical society: series B (methodological), 39(1):1–22, 1977

    1977

Show all 40 references
  1. [9]

    Heterogeneity for the win: One-shot federated clustering

    Don Kurian Dennis, Tian Li, and Virginia Smith. Heterogeneity for the win: One-shot federated clustering. In International Conference on Machine Learning, pages 2611–2620. PMLR, 2021

    2021

  2. [10]

    Concept decompositions for large sparse text data using clustering

    Inderjit S Dhillon and Dharmendra S Modha. Concept decompositions for large sparse text data using clustering. Machine learning, 42(1):143–175, 2001

    2001

  3. [11]

    Federated-em with heterogeneity mitigation and variance reduction

    Aymeric Dieuleveut, Gersende Fort, Eric Moulines, and Geneviève Robin. Federated-em with heterogeneity mitigation and variance reduction. Advances in Neural Information Processing Systems, 34:29553–29566, 2021

    2021

  4. [12]

    Flexible clustered federated learning for client-level data distribution shift

    Moming Duan, Duo Liu, Xinyuan Ji, Yu Wu, Liang Liang, Xianzhang Chen, Yujuan Tan, and Ao Ren. Flexible clustered federated learning for client-level data distribution shift. IEEE Transactions on Parallel and Distributed Systems, 33(11):2661–2674, 2021

    2021

  5. [13]

    Lane, Marc Langheinrich, and Mar- tin Gjoreski

    Dario Fenoglio, Mohan Li, Pietro Barbiero, Nicholas D. Lane, Marc Langheinrich, and Mar- tin Gjoreski. Flux: Efficient descriptor-driven clustered federated learning under arbitrary distribution shifts. In Advances in Neural Information Processing Systems (NeurIPS), 2025

    2025

  6. [14]

    An efficient framework for clustered federated learning

    Avishek Ghosh, Jichan Chung, Dong Yin, and Kannan Ramchandran. An efficient framework for clustered federated learning. Advances in neural information processing systems, 33:19586– 19597, 2020. 10

    2020

  7. [15]

    Robust federated learning in a heterogeneous environment

    Avishek Ghosh, Justin Hong, Dong Yin, and Kannan Ramchandran. Robust federated learning in a heterogeneous environment. arXiv:1906.06629, 2019

    1906 arXiv

  8. [16]

    Deep neural networks with random gaussian weights: A universal classification strategy? IEEE Transactions on Signal Processing, 64(13):3444–3457, 2016

    Raja Giryes, Guillermo Sapiro, and Alex M Bronstein. Deep neural networks with random gaussian weights: A universal classification strategy? IEEE Transactions on Signal Processing, 64(13):3444–3457, 2016

    2016

  9. [17]

    Delving deep into rectifiers: Surpassing human-level performance on imagenet classification

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Delving deep into rectifiers: Surpassing human-level performance on imagenet classification. In Proceedings of the IEEE international conference on computer vision, pages 1026–1034, 2015

    2015

  10. [18]

    Advances and open problems in federated learning

    Peter Kairouz and H Brendan McMahan. Advances and open problems in federated learning. Foundations and trends in machine learning, 14(1-2):1–210, 2021

    2021

  11. [19]

    Federated hyperparameter tuning: Challenges, baselines, and connections to weight-sharing

    Mikhail Khodak, Renbo Tu, Tian Li, Liam Li, Maria-Florina F Balcan, Virginia Smith, and Ameet Talwalkar. Federated hyperparameter tuning: Challenges, baselines, and connections to weight-sharing. Advances in Neural Information Processing Systems, 34:19184–19197, 2021

    2021

  12. [20]

    Recent advances on federated learning: A systematic survey

    Bingyan Liu, Nuoyan Lv, Yuanchun Guo, and Yawen Li. Recent advances on federated learning: A systematic survey. Neurocomputing, 597:128019, 2024

    2024

  13. [21]

    T-friedman test: A new statistical test for multiple comparison with an adjustable conservativeness measure

    Jie Liu and Yubo Xu. T-friedman test: A new statistical test for multiple comparison with an adjustable conservativeness measure. International Journal of Computational Intelligence Systems, 15(1):29, 2022

    2022

  14. [22]

    Multi-center federated learning: clients clustering for better personalization

    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, 26(1):481– 500, 2023

    2023

  15. [23]

    Privacy- preserving clustering federated learning for non-iid data

    Guixun Luo, Naiyue Chen, Jiahuan He, Bingwei Jin, Zhiyuan Zhang, and Yidong Li. Privacy- preserving clustering federated learning for non-iid data. Future Generation Computer Systems, 154:384–395, 2024

    2024

  16. [24]

    Structured federated learning through clustered additive modeling

    Jie Ma, Tianyi Zhou, Guodong Long, Jing Jiang, and Chengqi Zhang. Structured federated learning through clustered additive modeling. Advances in Neural Information Processing Systems, 36:43097–43107, 2023

    2023

  17. [25]

    Differentially private clustered federated learning

    Saber Malekmohammadi, Afaf Taik, and Golnoosh Farnadi. Differentially private clustered federated learning. Transactions on Machine Learning Research, 2025

    2025

  18. [26]

    Communication-efficient learning of deep networks from decentralized data

    Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, and Blaise Aguera y Arcas. Communication-efficient learning of deep networks from decentralized data. In Artificial intelligence and statistics, pages 1273–1282. PMLR, 2017

    2017

  19. [27]

    A survey on security and privacy of federated learning

    Viraaji Mothukuri, Reza M Parizi, Seyedamin Pouriyeh, Yan Huang, Ali Dehghantanha, and Gautam Srivastava. A survey on security and privacy of federated learning. Future Generation Computer Systems, 115:619–640, 2021

    2021

  20. [28]

    Machine learning: a probabilistic perspective

    Kevin P Murphy. Machine learning: a probabilistic perspective. MIT press, 2012

    2012

  21. [29]

    Fedshe: privacy pre- serving and efficient federated learning with adaptive segmented ckks homomorphic encryption

    Yao Pan, Zheng Chao, Wang He, Yang Jing, Li Hongjia, and Wang Liming. Fedshe: privacy pre- serving and efficient federated learning with adaptive segmented ckks homomorphic encryption. Cybersecurity, 7(1):40, 2024

    2024

  22. [30]

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

    Felix Sattler, Klaus-Robert Müller, and Wojciech Samek. Clustered federated learning: Model- agnostic distributed multitask optimization under privacy constraints. IEEE transactions on neural networks and learning systems, 32(8):3710–3722, 2020

    2020

  23. [31]

    Contrastive encoder pre-training-based clustered federated learning for heterogeneous data

    Ye Lin Tun, Minh NH Nguyen, Chu Myaet Thwal, Jinwoo Choi, and Choong Seon Hong. Contrastive encoder pre-training-based clustered federated learning for heterogeneous data. Neural Networks, 165:689–704, 2023. 11

    2023

  24. [32]

    Efficient distribution similarity identification in clustered federated learning via principal angles between client data subspaces

    Saeed Vahidian, Mahdi Morafah, Weijia Wang, Vyacheslav Kungurtsev, Chen Chen, Mubarak Shah, and Bill Lin. Efficient distribution similarity identification in clustered federated learning via principal angles between client data subspaces. In Proceedings of the AAAI conference ...

    2023

  25. [33]

    Fvfl: A flexible and verifiable privacy-preserving federated learning scheme

    Gang Wang, Li Zhou, Qingming Li, Xiaoran Yan, Ximeng Liu, and Yuncheng Wu. Fvfl: A flexible and verifiable privacy-preserving federated learning scheme. IEEE Internet of Things Journal, 11(13):23268–23281, 2024

    2024

  26. [34]

    Priver- ifl: Privacy-preserving and aggregation-verifiable federated learning

    Lulu Wang, Mirko Polato, Alessandro Brighente, Mauro Conti, Lei Zhang, and Lin Xu. Priver- ifl: Privacy-preserving and aggregation-verifiable federated learning. IEEE Transactions on Services Computing, 18(2):998–1011, 2024

    2024

  27. [35]

    Medmnist v2-a large-scale lightweight benchmark for 2d and 3d biomedical image classification

    Jiancheng Yang, Rui Shi, Donglai Wei, Zequan Liu, Lin Zhao, Bilian Ke, Hanspeter Pfister, and Bingbing Ni. Medmnist v2-a large-scale lightweight benchmark for 2d and 3d biomedical image classification. Scientific data, 10(1):41, 2023

    2023

  28. [36]

    Heterogeneous federated learning: State-of-the-art and research challenges

    Mang Ye, Xiuwen Fang, Bo Du, Pong C Yuen, and Dacheng Tao. Heterogeneous federated learning: State-of-the-art and research challenges. ACM Computing Surveys, 56(3):1–44, 2023

    2023

  29. [37]

    Stocfl: A stochastically clustered federated learning framework for non-iid data with dynamic client participation

    Dun Zeng, Xiangjing Hu, Shiyu Liu, Yue Yu, Qifan Wang, and Zenglin Xu. Stocfl: A stochastically clustered federated learning framework for non-iid data with dynamic client participation. Neural Networks, 187:107278, 2025

    2025

  30. [38]

    Efficient clustering on encrypted data

    Mengyu Zhang, Long Wang, Xiaoping Zhang, Zhuotao Liu, Yisong Wang, and Han Bao. Efficient clustering on encrypted data. In International Conference on Applied Cryptography and Network Security, pages 213–236. Springer, 2024

    2024

  31. [39]

    Em algorithms for gaussian mixtures with split-and-merge operation

    Zhihua Zhang, Chibiao Chen, Jian Sun, and Kap Luk Chan. Em algorithms for gaussian mixtures with split-and-merge operation. Pattern recognition, 36(9):1973–1983, 2003. A Experimental Details and Additional Results Experimental Setup.We consider 100 clients, each initially hold...

    1973

  32. [40]

    While FLAMECHE is not always the top-performing method in every individual configuration, it consistently achieves strong performance across all heterogeneity types. Table 4: Clustering Performance under Concept Shift on Labels MNIST Fashion-MNIST CIFAR-10 TissueMNIST PathMNIS...

Pith tools

Reviewed July 31, 2026 · model on record in the stance chip above.