Pith. sign in

REVIEW 3 major objections 5 minor 33 references

Adaptive Client Selection with Personalization for Communication Efficient Federated Learning

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read ACSP-FL shows that adapting the number and choice of federated-learning clients and sharing only part of each model can cut communication by up to 99% while keeping or raising accuracy on activity-recognition data.

desk verdict The communication savings are probably real; the accuracy advantage is not yet supported by the asymmetric evaluation. read the letter →

arxiv 2411.17833 v1 pith:AJ4A2ROZ submitted 2024-11-26 cs.LG cs.DC

Allan M. de Souza , Filipe Maciel , Joahannes B. D. da Costa , Luiz F. Bittencourt , Eduardo Cerqueira , Antonio A. F. Loureiro , Leandro A. Villas This is my paper · ORCID
classification cs.LGcs.DC
keywords federatedlearningclientselectionmodelpersonalizationpartialsharingcommunicationefficiencyhumanactivityrecognitionnon-IIDdata
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

ACSP-FL is a federated learning scheme built around one idea: the clients that need training most are the ones whose personalized models perform worst, and the network should only hear from those clients, in shrinking numbers, and never about the whole model. The paper argues that this triple mechanism—adaptive client selection, a decay that prunes the participant set as the model converges, and partial model sharing with client-side personalization—cuts the bytes exchanged between server and devices by up to 99% compared to standard federated averaging, on three human activity recognition datasets, while the distributed accuracy is as good as or better than the baselines. That matters because communication is the main bottleneck in federated learning: if the number of participating devices and the model size can both shrink during training without hurting accuracy, federated learning becomes practical on bandwidth-limited mobile networks.

What carries the argument

The argument runs on three coupled mechanisms. The selection filter $\pi(i, \mathcal{A})$ (Equation 4) admits a client only if its accuracy $\mathcal{A}_i$ is at or below the mean accuracy of all clients; the decay function $\phi(\mathcal{S}, t) = \lceil |\mathcal{S}| \cdot (1 - \text{decay})^t \rceil$ (Equation 6) then keeps only the first $\phi(\mathcal{S},t)$ of those below-average clients, so the number of participants falls as the model improves. Communication is cut by the layer-sharing operator $\mathcal{S}(w, L)$, which sends only the chosen layers of the model in both directions, and by the dynamic rule (Equation 9) that shares all layers only while accuracy is low and progressively shares fewer layers as a client's accuracy rises. Finally, the personalization function $\mathcal{I}(w_i^l, w^g)$ (Equation 8) lets each client keep the better of its fully local model and the global model at inference, which the paper argues is what keeps accuracy high even though most of the model is trained locally. Together these form a loop: better per-client models raise the mean accuracy, which shrinks the selected set and the number of shared layers, which in turn cuts bytes and latency.

What would settle it

Re-run the benchmarks evaluating every method two ways—once on the global model alone and once with the same best-of-local/global personalization applied to all baselines; if ACSP-FL's accuracy advantage disappears or reverses when baselines get the personalization, then the claimed convergence gain is an artifact of the evaluation protocol rather than of the adaptive client selection.

Watch

Extended reading notes

Core claim

The central claim is that ACSP-FL, by coupling performance-based client selection with model personalization and layer sharing, achieves communication-efficient convergence without the accuracy loss usually associated with aggressive communication reduction. Selection is driven by each round's distributed evaluation: clients whose accuracy falls at or below the mean become the training set for the next round, and a decay function then shrinks that set further as rounds progress, so participation drops automatically at convergence instead of staying fixed. In parallel, each client keeps a fully local model and receives only the shared layers of the global model; at evaluation time it keeps whichever of the two—local or global—performs better on its own data, and the number of shared layers is reduced as client accuracy rises. On the UCI-HAR, MotionSense, and ExtraSensory datasets, ACSP-FL is reported to send roughly 99%, 99%, and 97% fewer bytes than FedAvg respectively, cut latency by about 90%, and achieve higher distributed accuracy, including on the non-IID ExtraSensory data.

Load-bearing premise

The accuracy comparison is asymmetric: ACSP-FL clients pick the better of their local and global model at evaluation time, while FedAvg, POC, OORT, and DEEV are judged only on the shared global model, so if the baselines were given the same personalization the gap might shrink or vanish.

Editorial extensions

If this is right

  • If the reported reductions hold, federated learning can run with roughly 1% of its usual uplink and downlink traffic on the same model architecture, with no fixed per-round client count to tune.
  • The automatic shrink of both the participant set and the shared layer count means the scheme needs no k-value or compression schedule set in advance, which removes a common hyperparameter burden for deployment.
  • Because clients with above-average personalized accuracy stop being called to train, devices save both bandwidth and compute, which directly translates into longer battery life and lower network cost in mobile settings.
  • The same or better accuracy is reported on a non-IID benchmark (ExtraSensory), suggesting the approach is not limited to identically distributed client data.

Reading between the lines

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

  • Because ACSP-FL's selection rule reacts to personalized client accuracy, the natural ablation—running the same pipeline without the Equation 8 personalization—would separate how much of the byte reduction comes from adaptive selection and layer sharing versus from the accuracy boost that personalization itself provides.
  • The selection loop it creates (personalization raises accuracy, which shrinks the below-average set, which cuts bytes further) leaves open whether the chosen clients stay representative on strongly heterogeneous data; logging the data distribution of selected clients across rounds would test this.
  • Each client now maintains both a local and a global model and runs an extra evaluation pass per round, so the scheme trades per-device storage and computation for communication; reporting those costs alongside the byte savings would show where the tradeoff lands.
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. ACSP-FL combines adaptive client selection (based on client accuracy relative to the mean), a decay function that reduces the number of participating clients over rounds, and partial model sharing with per-client personalization. The authors evaluate it on three HAR datasets (UCI-HAR, MotionSense, ExtraSensory) against FedAvg, POC, OORT, and their own DEEV, claiming communication reductions up to 95-99%, faster convergence, and higher accuracy. The communication-reduction direction is plausible and supported by TX-byte measurements, but the accuracy and convergence claims rest on an asymmetric evaluation protocol in which ACSP-FL uses personalized per-client models while the baselines are evaluated as a single global model.

Significance. If the reported communication reductions hold under a fair evaluation, the contribution is practically useful: it directly attacks the two dominant costs in federated learning (round count and per-round bytes), and the paper ships open-source code, a container-based testbed, and measurements on three real datasets rather than only synthetic ones. The TX-byte accounting is a strength. However, the headline accuracy and convergence claims are currently not supported because the comparison confounds personalization with improved shared-model convergence; the significance is therefore conditional on the authors providing a controlled personalization baseline and on resolving the suspiciously high initial accuracy values in the learning curves.

major comments (3)
  1. [Section 4.3 and Table 4] The distributed-accuracy metric is computed differently for ACSP-FL and the baselines. For ACSP-FL, Algorithm 2 (Evaluate) constructs each client model as w_i = [w_g, w_l_i], and Eq. (8) lets the client select whichever of the local composite model and the global model has lower loss on its own data; for FedAvg, POC, OORT, and DEEV, the single global model is evaluated without fine-tuning or local-model selection. Consequently, the higher accuracies in Table 4 and Figure 8 confound personalization with better convergence of the shared model. Please add a controlled personalization baseline (e.g., per-client fine-tuning of the global model or the same local/global selection rule for all methods) and also report the accuracy of the pure shared global model for ACSP-FL.
  2. [Figures 6 and 8; Section 4.1] The accuracy curves start at implausibly high values at round 0. For example, on ExtraSensory the first plotted point shows FedAvg around 0.84 and ACSP-FL DLD around 0.92, although Algorithm 1 initializes the model randomly (RANDOMINIT) and a randomly initialized MLP should give near-chance accuracy on a multiclass HAR task. Please clarify whether the x-axis starts at the initial random model, and state the local training hyperparameters (number of epochs, learning rate, batch size, optimizer settings). If the models are warm-started or pre-trained, this must be disclosed, since it changes the interpretation of both the convergence and communication-savings claims.
  3. [Tables 3-4 and Section 4.5] All reported results are single runs, and both the ACSP-FL hyperparameters (decay = 0.005 in Section 4.5; the PMS mapping in Eq. (9)) and the baseline hyperparameters (k = 50% for OORT and POC) were chosen via exploratory analysis without stating whether the tuning was done on test data. Without multiple seeds, error bars, or a sensitivity analysis for decay and the layer-sharing threshold, the magnitude of the claimed reductions (e.g., 97-99% fewer TX bytes versus FedAvg) cannot be assessed for robustness. Please report means and standard deviations over at least three runs and describe how the exploratory hyperparameter selection is separated from the reported test results.
minor comments (5)
  1. [Abstract, Section 4.5, Conclusion] The reported savings are inconsistent: the abstract and conclusion claim up to 95% communication reduction, while Table 4 shows 97-99% TX-byte reductions versus FedAvg; similarly, Section 4.5 says accuracy increases up to 8%, while the conclusion says at least 10%. Please reconcile these numbers.
  2. [Equation (9) and Section 4.2] The dynamic layer-sharing function states that '4 is the total number of layers of the model' but the MLP is described as having three hidden layers; clarify which layers are counted (input, hidden, output) and explain the behavior of the formula at the accuracy = 0.25 threshold, where ceil(1/0.25) = 4 yields no change in shared layers.
  3. [Figure 5] The x-axis of Figure 5 is labeled 'Client Accuracy (%)' but the plotted values range from 0.0 to 1.0; use fraction or percent consistently to avoid confusion.
  4. [Algorithm 2, line 2] The comment in Algorithm 2 says 'combines the local model with the local one'; this should read 'combines the global piece with the local piece' to match the w_i = [w_g, w_l_i] construction.
  5. [References [33] and [34]] The same paper by Vaizman et al. appears as both reference [33] and reference [34]; the duplicate should be merged.

Circularity Check

0 steps flagged · score 2.0 of 10

No derivation-level circularity: communication savings are measured, and the accuracy comparison is asymmetric but not forced by construction.

full rationale

The paper's main claims—communication reduction and accuracy preservation/improvement—are supported by direct measurements on public HAR datasets (UCI-HAR, MotionSense, ExtraSensory), not by fitting a parameter to a subset and then predicting a closely related quantity. The client-selection equations (Eqs. 4–7) and the dynamic layer-sharing function (Eq. 9) are heuristics that define the mechanism, not predictions derived tautologically from their own outputs. The adaptive selection uses the same accuracy metric as the evaluation, but that is a feedback mechanism in the algorithm, not a circular proof of the algorithm's performance. The paper transparently extends the authors' prior DEEV work [9] and evaluates DEEV as a baseline; no load-bearing argument reduces to an unverified self-citation. The efficiency metric is author-defined with alpha=beta=0.5, but it is a reporting metric rather than a predicted result. The most significant weakness is the asymmetric evaluation protocol: ACSP-FL clients evaluate a personalized model (Algorithm 2 combines shared and local layers, and Section 3.4 allows choosing the better of local and global models), while FedAvg, OORT, POC, and DEEV are evaluated only as a single global model. This is a fairness concern about the accuracy comparison, but it is an experimental-design issue, not a circular derivation: the reported accuracy gains are empirical and could in principle have gone the other way if local layers overfit. Overall, no circular step meets the evidence bar of quoting an equation that reduces to its inputs.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claims rest on empirical system evaluation rather than a formal derivation. The free parameters are evaluation hyperparameters; the axioms are assumptions about FL convergence and the reliability of accuracy-based selection.

free parameters (3)
  • decay = 0.005
    Controls exponential client-count reduction in Eq. 6; set for the evaluation, not derived.
  • PMS dynamic-layer threshold and mapping = 0.25; ceil(1/accuracy)
    Eq. 9 uses an ad hoc threshold and inverse-accuracy rule to decide how many of the 4 layers are shared.
  • Efficiency metric weights alpha and beta = 0.5 and 0.5
    The reported efficiency score is 0.5*accuracy + 0.5*overhead reduction; changing weights changes all efficiency claims.
assumptions (4)
  • domain assumption Federated averaging with partial client participation converges under the ACSP-FL selection and decay schedule
    Section 3.1 defers convergence of partial client selection to [7] and provides no convergence analysis for the added personalization, layer sharing, and decay.
  • domain assumption Client-reported test accuracy is a reliable selection signal
    Eq. 4 selects clients whose accuracy is below the mean; the paper does not analyze whether this selection accelerates convergence or is robust to noisy accuracy reports.
  • ad hoc to paper Sharing a subset of layers preserves model quality
    The dynamic layer-sharing formula in Eq. 9 is a heuristic with no error bound or convergence guarantee.
  • domain assumption Personalized local models do not overfit
    Section 4.5 asserts ACSP-FL does not overfit, but no regularization, generalization bound, or analysis of local-model drift is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Adaptive Client Selection with Personalization for Communication Efficient Federated Learning." pith.science (2026). https://pith.science/paper/AJ4A2ROZ

@misc{pith2026241117833,
  author       = {Pith},
  title        = {Pith review of: Adaptive Client Selection with Personalization for Communication Efficient Federated Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AJ4A2ROZ}},
  note         = {Machine review of arXiv:2411.17833}
}
read the original abstract

Federated Learning (FL) is a distributed approach to collaboratively training machine learning models. FL requires a high level of communication between the devices and a central server, thus imposing several challenges, including communication bottlenecks and network scalability. This article introduces ACSP-FL (https://github.com/AllanMSouza/ACSP-FL), a solution to reduce the overall communication and computation costs for training a model in FL environments. ACSP-FL employs a client selection strategy that dynamically adapts the number of devices training the model and the number of rounds required to achieve convergence. Moreover, ACSP-FL enables model personalization to improve clients performance. A use case based on human activity recognition datasets aims to show the impact and benefits of ACSP-FL when compared to state-of-the-art approaches. Experimental evaluations show that ACSP-FL minimizes the overall communication and computation overheads to train a model and converges the system efficiently. In particular, ACSP-FL reduces communication up to 95% compared to literature approaches while providing good convergence even in scenarios where data is distributed differently, non-independent and identical way between client devices.

Figures

Figures reproduced from arXiv: 2411.17833 by the authors.

Figure 1
Figure 1. Performance-based client selection with personalization: personalization phase (a) the server sends the model to the clients, the portion of the model is defined by (⋅); in evaluation phase (b) clients combine the layers shared by the server with their local model to train it with their local data and send back the performance metric to the server; selection phase (c), the server classifies clients according to per… view at source ↗
Figure 2
Figure 2. Example of the decay mechanism employed by ACSP￾FL. Allan M. de Souza et al.: Preprint submitted to Elsevier Page 5 of 16 [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. shows how ACSP-FL builds the model in each client combining the shared layers (i.e., the global piece 𝑤𝑔 ) with the local layers personalized in each client (i.e., the local piece 𝑤𝑙 𝑖 ). In this example, 𝐿 = {𝑙1 , 𝑙2 } defines that the first two layers, the black neurons, are the shared global model 𝑤𝑔 = (𝑤, 𝐿), while the remaining part, the red neurons, will be personalized on its local data. Therefore, combining… view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Visualization of classes distribution by client for each dataset. • Efficiency: this metric evaluates the efficiency of the solution to reach the provided accuracy by consider￾ing a weighted sum of the average accuracy and the overhead reduction. In particular, the wei…
Figure 5
Figure 5. Figure 5: Example of dynamic layer definition function based on client’s accuracy. The results indicate that, despite the number of layers shared, ACSP-FL can still converge for each dataset (refer to [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Test accuracy in function of the communication round for each dataset. 20 40 60 80 100 Communication Round 10 3 10 4 10 5 10 6 10 7 TX Data (Bytes) UCIHAR 20 40 60 80 100 Communication Round 10 3 10 4 10 5 10 6 MotionSense FedAvg ACSP-FL ND ACSP-FL FT ACSP-FL PMS 3 ACS…
Figure 7
Figure 7. Figure 7: Analysis of communication overhead considering the partial model sharing and also the decaying function be a suitable solution, as no hyperparameter exploration is needed for different scenarios. The performance of ACSP-FL variants for other metrics is summarized in […
Figure 8
Figure 8. Figure 8: Test accuracy in function of the communication round for each dataset. each round. Furthermore, for ACSP-FL, we set the value for 𝑑𝑒𝑐𝑎𝑦 = 0.005. As seen in [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Latency produced by each solution for each dataset [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: Accuracy distribution for each client at the last communication round. may hide the performance of specific clients with notably low accuracy. Therefore, when considering the number of clients higher than the average accuracy, we can see that for ACSP￾FL, more clients…
Figure 11
Figure 11. Figure 11: Evaluation of how many times clients were selected by each solution. When examining ACSP-FL with adaptive client selec￾tion, personalization, and layer sharing, clients are chosen less frequently for model training. This is a consequence of improved client performance…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 23 canonical work pages

  1. [1]

    P.Porambage,J.Okwuibe,M.Liyanage,M.Ylianttila,T.Taleb, Sur- veyonmulti-accessedgecomputingforinternetofthingsrealization, IEEE Communications Surveys & Tutorials 20 (2018) 2961–2991

    work page 2018

  2. [2]

    N.Abbas,Y.Zhang,A.Taherkordi,T.Skeie,Mobileedgecomputing: A survey, IEEE Internet of Things Journal 5 (2017) 450–465

    work page 2017

  3. [3]

    W. Y. B. Lim, N. C. Luong, D. T. Hoang, Y. Jiao, Y.-C. Liang, Q. Yang, D. Niyato, C. Miao, Federated learning in mobile edge networks: A comprehensive survey, IEEE Communications Surveys & Tutorials 22 (2020) 2031–2063

    work page 2020

  4. [4]

    AbdulRahman, H

    S. AbdulRahman, H. Tout, H. Ould-Slimane, A. Mourad, C. Talhi, M. Guizani, A survey on federated learning: The journey from centralized to distributed on-site learning and beyond, IEEE Internet of Things Journal 8 (2020) 5476–5497

    work page 2020

  5. [5]

    Zhang, X

    X. Zhang, X. Zhu, J. Wang, H. Yan, H. Chen, W. Bao, Federated learning with adaptive communication compression under dynamic bandwidthandunreliablenetworks, InformationSciences540(2020) 242–262

    work page 2020

  6. [6]

    O. A. Wahab, A. Mourad, H. Otrok, T. Taleb, Federated machine learning: Survey, multi-level classification, desirable criteria and fu- ture directions in communication and networking systems, IEEE Communications Surveys & Tutorials 23 (2021) 1342–1397

    work page 2021

  7. [7]

    Y. J. Cho, J. Wang, G. Joshi, Client selection in federated learning: Convergenceanalysisandpower-of-choiceselectionstrategies, arXiv preprint arXiv:2010.01243 (2020)

    arXiv 2020

  8. [8]

    A. Z. Tan, H. Yu, L. Cui, Q. Yang, Towards personalized federated learning, IEEE Transactions on Neural Networks and Learning Systems (2022) 1–17

    work page 2022

Show all 33 references
  1. [9]

    A. M. de Souza, L. Bittencourt, E. Cerqueira, A. Loureiro, L. Vil- las, Dispositivos, eu escolho vocês: Seleção de clientes adaptativa para comunicação eficiente em aprendizado federado, in: Anais do XLI Simpósio Brasileiro de Redes de Computadores e Sistemas Distribuídos, SBC...

    2023

  2. [10]

    McMahan, E

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

    2017

  3. [11]

    S. M. Shah, V. K. N. Lau, Model compression for communication efficient federated learning, IEEE Transactions on Neural Networks and Learning Systems 34 (2023) 5937–5951. 3https://github.com/AllanMSouza/ACSP-FL

    2023

  4. [12]

    Y. Lin, S. Han, H. Mao, Y. Wang, B. Dally, Deep gradient compres- sion:Reducingthecommunicationbandwidthfordistributedtraining, in: International Conference on Learning Representations, 2018, pp. 1–14. URL:https://openreview.net/pdf?id=SkhQHMW0W

    2018

  5. [13]

    M. M. Amiri, D. Gunduz, S. R. Kulkarni, H. V. Poor, Feder- ated learning with quantized global model updates, arXiv preprint arXiv:2006.10672 (2020)

    2020 arXiv

  6. [14]

    Bernstein, Y.-X

    J. Bernstein, Y.-X. Wang, K. Azizzadenesheli, A. Anandkumar, signsgd: Compressed optimisation for non-convex problems, in: International Conference on Machine Learning, PMLR, 2018, pp. 560–569

    2018

  7. [15]

    Sattler, S

    F. Sattler, S. Wiedemann, K.-R. Müller, W. Samek, Robust and communication-efficientfederatedlearningfromnon-i.i.d.data,IEEE Transactions on Neural Networks and Learning Systems 31 (2020) 3400–3413

    2020

  8. [16]

    Ström, Scalable distributed dnn training using commodity gpu cloud computing, in: Interspeech 2015, 2015, pp

    N. Ström, Scalable distributed dnn training using commodity gpu cloud computing, in: Interspeech 2015, 2015, pp. 1–5. URL: https://www.amazon.science/publications/ scalable-distributed-dnn-training-using-commodity-gpu-cloud-computing

    2015

  9. [17]

    Shahid, S

    O. Shahid, S. Pouriyeh, R. M. Parizi, Q. Z. Sheng, G. Srivastava, L. Zhao, Communication efficiency in federated learning: Achieve- ments and challenges, arXiv preprint arXiv:2107.10996 (2021)

    2021 arXiv

  10. [18]

    G. D. Németh, M. A. Lozano, N. Quadrianto, N. M. Oliver, A snapshot of the frontiers of client selection in federated learning, Transactions on Machine Learning Research (2022)

    2022

  11. [19]

    Nishio, R

    T. Nishio, R. Yonetani, Client selection for federated learning with heterogeneous resources in mobile edge, in: ICC 2019 - 2019 IEEE International Conference on Communications (ICC), 2019, pp. 1–7. doi:10.1109/ICC.2019.8761315

    2019

  12. [20]

    Goetz, K

    J. Goetz, K. Malik, D. Bui, S. Moon, H. Liu, A. Kumar, Active federated learning, arXiv preprint arXiv:1909.12641 (2019)

    2019 arXiv

  13. [21]

    J. Kang, Z. Xiong, D. Niyato, S. Xie, J. Zhang, Incentive mechanism for reliable federated learning: A joint optimization approach to combining reputation and contract theory, IEEE Internet of Things Journal 6 (2019) 10700–10714

    2019

  14. [22]

    F. Lai, X. Zhu, H. V. Madhyastha, M. Chowdhury, Oort: Effi- cient federated learning via guided participant selection, in: 15th USENIXSymposiumonOperatingSystemsDesignandImplementa- tion(OSDI21),USENIXAssociation,2021,pp.19–35.URL: https: //www.usenix.org/conference/osdi21/pre...

    2021

  15. [23]

    H. Chen, J. Ding, E. W. Tramel, S. Wu, A. K. Sahu, S. Avestimehr, T. Zhang, Self-aware personalized federated learning, in: Advances in Neural Information Processing Systems, 2022, pp. 20675–20688

    2022

  16. [24]

    Kairouz, H

    P. Kairouz, H. B. McMahan, B. Avent, A. Bellet, M. Bennis, A. N. Bhagoji, K. Bonawitz, Z. Charles, G. Cormode, R. Cummings, et al., Advances and open problems in federated learning, Foundations and Trends®in Machine Learning 14 (2021) 1–210

    2021

  17. [25]

    B. Luo, X. Li, S. Wang, J. Huang, L. Tassiulas, Cost-effective federated learning design, in: IEEE INFOCOM 2021 - IEEE Con- ference on Computer Communications, 2021, pp. 1–10. doi:10.1109/ INFOCOM42981.2021.9488679

    2021

  18. [26]

    A. H. Mohamed, N. R. G. Assumpçáo, C. A. Astudillo, A. M. deSouza,L.F.Bittencourt,L.A.Villas, Compressedclientselection for efficient communication in federated learning, in: 2023 IEEE 20thConsumerCommunications&NetworkingConference(CCNC), 2023, pp. 508–516. doi:10.1109/CCNC51...

    2023

  19. [28]

    C.G.S.Capanema,A.M.deSouza,F.A.Silva,L.A.Villas,A.A.F. Loureiro, Fedpredict: Combining global and local parameters in the prediction step of federated learning, in: 2023 19th Interna- tional Conference on Distributed Computing in Smart Systems and the Internet of Things (DCOSS...

    2023

  20. [29]

    Haykin, Neural Networks: A Comprehensive Foundation, 2nd ed., Prentice Hall PTR, USA, 1998

    S. Haykin, Neural Networks: A Comprehensive Foundation, 2nd ed., Prentice Hall PTR, USA, 1998

    1998

  21. [30]

    D. J. Beutel, T. Topal, A. Mathur, X. Qiu, J. Fernandez-Marques, Y. Gao, L. Sani, K. H. Li, T. Parcollet, P. P. B. de Gusmão, et al., Flower: A friendly federated learning research framework, arXiv preprint arXiv:2007.14390 (2020)

    2020 arXiv

  22. [31]

    Reyes-Ortiz, D

    J. Reyes-Ortiz, D. Anguita, A. Ghio, L. Oneto, X. Parra, Human Activity Recognition Using Smartphones, UCI Machine Learning Repository, 2012. DOI: https://doi.org/10.24432/C54S4K

    2012 doi

  23. [32]

    Malekzadeh, R

    M. Malekzadeh, R. G. Clegg, A. Cavallaro, H. Haddadi, Mobile sensor data anonymization, in: Proceedings of the International Conference on Internet of Things Design and Implementation, As- sociation for Computing Machinery, 2019, p. 49–58. URL:https: //doi.org/10.1145/3302505....

    2019

  24. [33]

    Vaizman, K

    Y. Vaizman, K. Ellis, G. Lanckriet, Recognizing detailed human context in the wild from smartphones and smartwatches, IEEE Pervasive Computing 16 (2017) 62–74

    2017

  25. [34]

    Vaizman, K

    Y. Vaizman, K. Ellis, G. Lanckriet, Recognizing detailed human context in the wild from smartphones and smartwatches, IEEE Pervasive Computing 16 (2017) 62–74. Allan M. de Souza et al.:Preprint submitted to Elsevier Page 16 of 16

    2017

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.