Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

FedCod: An Efficient Communication Protocol for Cross-Silo Federated Learning with Coding

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read FedCod encodes model updates into redundant blocks and relays them over idle client-to-client links, cutting total cross-silo federated learning communication time by up to 62% relative to direct server-client transfer while preserving…

desk verdict Solid systems paper with a directly measured 62% communication cut under a clearly stated assumption about idle inter-client bandwidth; the main weaknesses are evaluation rigor, weighted-aggregation compatibility, and no sensitivity to restricted inter-client links. read the letter →

arxiv 2501.00216 v1 pith:4HAIYDTS submitted 2024-12-31 cs.DC

Peishen Yan , Jun Li , Hao Wang , Tao Song , Yang Hua , Lu Peng , Haihui Zhou , Haibing Guan This is my paper · ORCID
classification cs.DC
keywords federatedlearningcross-silocommunicationprotocolnetworkcodingrandomlinearadaptiveredundancyclient-to-clientefficiency
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

FedCod is an application-layer communication protocol for cross-silo federated learning (FL), where organizations train a shared model without sharing data. The paper claims that the slow, fluctuating wide-area links between silos and the server are the main bottleneck, and that this bottleneck can be bypassed rather than accepted or worked around algorithmically. FedCod splits models into partitions, encodes them into redundant linear-combination blocks, and routes those blocks over idle client-to-client links, so clients can reconstruct the global model from whichever blocks arrive first. Measured on real cloud deployments across a global topology and a North America topology, FedCod reports a 59% reduction in average download time and a 62% reduction in total communication time versus the baseline, with convergence and final accuracy unchanged. If these numbers hold, cross-silo FL can become much faster without changing the learning algorithm, losing information, or adding a centralized fast network.

What carries the argument

The central object is random linear coding over model partitions, applied at the application layer with a forwarding rule that only relays blocks received directly from the server, never re-encoding or re-forwarding received blocks, to avoid linear dependence. Coded-AGR uses a pre-agreed Cauchy-matrix coefficient sequence so clients can sum blocks with the same coefficient vector, reducing server ingress traffic. The adaptive redundancy algorithm maintains a lower bound on redundancy and adjusts it based on comparisons of current versus previous round durations scaled by a factor $\lambda > 1$, starting high, decreasing while performance is stable, and recovering quickly when fluctuations are detected.

What would settle it

Deploy FedCod on a testbed where every client-to-client link is capped to near-zero spare bandwidth while server-client links remain unchanged; if total communication time stays close to the baseline, the client-to-client forwarding path is what carries the 62% reduction.

Watch

Extended reading notes

Core claim

At the core of FedCod is the claim that communication efficiency in cross-silo FL can be improved losslessly at the application layer, independent of the training algorithm. The protocol partitions each model into $k$ equal pieces, forms encoded blocks as random linear combinations $M_t^i = G_t \cdot A_t^i$ with coefficient vectors $A_t^i$, and distributes these blocks so that clients forward what they receive to neighbors; once a client has any $k$ linearly independent blocks it decodes the full model by Gaussian elimination. In the upload direction, all clients use the same predetermined coefficient-vector sequence, and each relay client aggregates encoded blocks that share a coefficient vector into a Coded-AGR block before sending to the server, so the server decodes the aggregated global model rather than every individual local model. The protocol also includes an adaptive redundancy controller that starts with high redundancy, reduces it while communication times stay stable, and raises it again when fluctuations are detected. On real geo-distributed deployments with ten global silos the paper reports a 59% average download-time reduction and a 62% total communication-time reduction against the basic server-client baseline, with server ingress traffic in the wait-mode Coded-AGR upload dropped to about 11% of baseline, and identical test accuracy at convergence.

Load-bearing premise

The performance gain rests on there being idle, usable bandwidth on client-to-client links that can be borrowed without cost or policy barriers; if those links are unavailable, saturated, or blocked, the reported reductions do not materialize.

Editorial extensions

If this is right

  • Cross-silo FL deployments using FedCod can cut per-round communication time by more than half without changing the training algorithm or the final model accuracy.
  • The download phase shows the largest gains, a 59% average download-time reduction in the global topology, indicating that coded forwarding helps most when one source must reach many silos.
  • Wait-mode Coded-AGR cuts server ingress traffic to about 11% of baseline, so the protocol reduces both latency and server bandwidth cost in the upload phase.
  • Hierarchical aggregation is not a reliable fallback for geo-distributed silos: in the paper's global topology it performed worse than the simple server-client baseline.
  • Because the protocol is lossless, it preserves convergence: all tested configurations reached the same 0.80 test accuracy as the baseline.

Reading between the lines

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

  • The measured gains presume idle client-to-client links; in a deployment where inter-silo traffic is firewalled or fully utilized, FedCod's benefit should shrink, and the paper does not quantify that case.
  • The adaptive redundancy controller is a generic mechanism: the same start-high, decrease-when-stable, recover-on-fluctuation rule could be applied to other redundant transmission systems, though the paper does not claim this.
  • FedCod's linear coding could be composed with compression or sparsification of model updates, but the paper does not test whether the combined lossy-plus-lossless pipeline preserves the reported gains.
  • Because relay clients see coded linear combinations of other silos' model updates, deployments with strict data-sharing policies may need secure aggregation or differential privacy on top; the paper does not address this.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. FedCod is an application-layer communication protocol for cross-silo federated learning that encodes model weight partitions and forwards encoded blocks over client-to-client links to exploit idle bandwidth, with a Coded-AGR mechanism for on-client aggregation and an adaptive redundancy controller. The paper evaluates the protocol on AWS/Azure deployments in a global and a North America topology, reporting an up-to-62% reduction in total communication time, large server traffic savings, unchanged convergence of a ResNet152/CIFAR-10 FL task, and additional numerical experiments on partition count, redundancy, and faulty links.

Significance. If the central claim holds, FedCod would be a practically useful, lossless, FL-algorithm-agnostic communication layer for cross-silo FL: it measures communication time directly against a baseline rather than fitting the outcome, and it provides consistent traffic accounting across protocol variants. The paper also includes conformance experiments showing unchanged convergence and profile-based numerical experiments for parameter sensitivity, which are strengths. The significance is conditional on two load-bearing points: the availability of usable idle client-to-client bandwidth, and Coded-AGR actually implementing the server's weighted aggregation rule; both need explicit treatment before the headline reduction can be taken as a general cross-silo claim.

major comments (4)
  1. [Section III-B3, paragraph after Fig. 4] The Coded-AGR aggregation step computes \tilde M_j = \sum_i M^{t,i}_j, an unweighted sum of encoded client updates. The server then decodes \sum_i W^{t+1}_i from the AGR blocks. In FedAvg, however, the global update is the dataset-size weighted average W_{t+1} = \sum_i (|D_i|/\sum_j |D_j|) W^{t+1}_i, as stated in Section II-A. Unless all clients have identical dataset sizes, Coded-AGR does not implement FedAvg and is not FL-algorithm-agnostic in the claimed sense. Please either encode the weighted updates p_i W^{t+1}_i, restrict the compatibility claim to equal dataset sizes, or provide a weighted aggregation mechanism; the conformance experiments in Table III do not resolve this because the dataset partition sizes are not reported.
  2. [Section III-B3, Proposition 1 and Eq. (3)] The proof of Proposition 1 does not compare wait mode with non-wait mode as those modes are defined. The protocol defines non-wait mode as uploading an AGR block once a predefined time window has elapsed, which can occur before all j-th blocks arrive, and wait mode as uploading after all such blocks are received. In the proof, the 'non-wait' case sends the AGR block at t' after the last block has already arrived at the client, which is actually wait-mode behavior with a delayed start, and it implicitly assumes the AGR block contains all client blocks. The inequality in Eq. (3) therefore supports a different statement than Proposition 1. Since U3-AGR is the variant used in the final FEDCOD configuration, the theoretical justification should be corrected or replaced by the empirical U2-AGR versus U3-AGR comparison.
  3. [Section IV-A and IV-B1] The quantitative headline relies on point estimates without variance information. The setup runs 10 communication rounds per protocol over one bandwidth realization, and WAN bandwidth fluctuates substantially as shown in Figure 1; Figure 5 reports single bars with no confidence intervals or per-round distributions. In the North America topology, training time is sampled from a distribution rather than measured, so end-to-end waiting time there is partly synthetic. Please report per-round time distributions, standard errors or confidence intervals, and clarify whether the 62% figure is a mean over rounds or over a single run.
  4. [Sections II-C, III-B1, and III-B2] All gains depend on the availability of idle, usable client-to-client bandwidth, stated in Section II-C as 'there are unoccupied client-to-client links and idle bandwidth that could be utilized.' The evaluation only profiles cloud inter-region links (Figure 7); it does not test a network with restricted or absent silo-to-silo connectivity, which is common in real cross-silo FL due to firewalls or data-governance policy. In that case the forwarding and Coded-AGR mechanisms cannot operate and the reported reduction collapses. This is a scope limitation rather than an internal inconsistency, but it should be made explicit and tested, for example by simulating disabled or saturated client-to-client links and reporting the resulting communication time.
minor comments (5)
  1. [Section II-B] The text contains typos such as 'lenthy' and 'flcutuating'; a careful proofreading pass is needed.
  2. [Section III-C] The adaptive redundancy algorithm depends on the initial redundancy r, the scaling factor lambda, and the lower bound r_lb, but no sensitivity analysis is reported for these parameters; the claimed 11% and 25% traffic reductions are for one parameter setting. A short sensitivity study or explicit default-parameter rationale would strengthen the adaptive-redundancy claim.
  3. [Section IV-A] The default is described as 'we set the number of model partitions k to n', but Figure 8(b) tests upload partition counts up to 30 while the deployment has n=10 clients; please clarify whether those experiments vary the number of clients or treat local partition count independently.
  4. [Section IV-B4] The conformance experiments are performed on a local cluster, but the details of the data partition, number of clients, and whether dataset sizes are equal are not given; these details matter for interpreting the Coded-AGR convergence results.
  5. [General] No code or data artifacts are released, which limits reproducibility of the measured 62% reduction and of the profiling in Figures 1 and 7; making the implementation and traces available would be valuable.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the headline 62% communication-time reduction is a direct measurement against a baseline, not a fitted or self-referential quantity.

full rationale

The paper's central claim (Section IV-B1: "With the full application of our coding strategies, FEDCOD reduces the average download time by 59% and the total communication time by 62% compared to the baseline") is an empirical comparison of measured communication times on real cloud deployments. The coding strategies are defined constructively in Sections III-B1 and III-B2, and the adaptive redundancy algorithm in Section III-C is a feedback controller that adjusts redundancy based on observed communication duration; it is not fitted to the outcome being reported. The only analytical theorem, Proposition 1 (wait mode vs. non-wait mode Coded-AGR), is a comparison between two modes within the proposed design and is not used to derive the headline reduction. The numerical experiments in Section IV-B3 use profiled bandwidth as simulation inputs, but they explore parameter sensitivity rather than claiming to predict the 62% figure. The sole self-citation with author overlap is reference [44] (which includes co-author Hao Wang), used only to specify the CIFAR-10 federated dataset-generation procedure in the experimental setup; this is not load-bearing for any claimed derivation or performance result. No step in the paper reduces a reported prediction to a fitted parameter, a definitional identity, or a self-citation chain. Accordingly, no significant circularity is present.

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

The result rests on the availability and usability of inter-client bandwidth, linearity of FL aggregation, and coding-theoretic independence of random coefficient vectors. The adaptive redundancy algorithm introduces several hand-chosen parameters whose values are not fully disclosed. No invented entities are proposed.

free parameters (5)
  • Number of model partitions k = Set to n (number of clients) by default
    Controls coding granularity; experiments show k=n chosen by hand, with performance trade-offs.
  • Initial coding redundancy r = Large (not specified); static experiments use 100%
    Adaptive algorithm starts with high redundancy; the exact initial value is not disclosed.
  • Redundancy scaling factor lambda = Not specified
    Used to decide when to reduce or restore redundancy; exact value affects behavior.
  • Lower bound r_lb of redundancy = Adjusted heuristically; no exact schedule
    Sets minimum resilience; update rules are described qualitatively.
  • Non-wait upload time window = Not specified
    Determines when non-wait Coded-AGR uploads; never quantified.
assumptions (5)
  • standard math Random coefficient vectors produce linearly independent encoded blocks with high probability.
    Used in Section III-B1 for client-side decoding and forwarding.
  • domain assumption Client-to-client links have idle bandwidth available for forwarding at no extra cost.
    Motivated in Section II-C and used throughout the protocol design.
  • domain assumption Federated aggregation is linear and Coded-AGR can reproduce the exact aggregate, including weighted FedAvg.
    Invoked in Section III-B3; the paper only demonstrates simple sums, not weighting by dataset sizes.
  • ad hoc to paper Bandwidth from client to server is positive and queueing delay is negligible in the wait-mode comparison.
    Proposition 1 proof in Section III-B2 assumes these to conclude wait mode is better.
  • domain assumption Coding and decoding are lossless and do not change the model update values.
    Section IV-B4 asserts the protocol is lossless; relies on exact linear algebra over reals.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FedCod: An Efficient Communication Protocol for Cross-Silo Federated Learning with Coding." pith.science (2026). https://pith.science/paper/4HAIYDTS

@misc{pith2026250100216,
  author       = {Pith},
  title        = {Pith review of: FedCod: An Efficient Communication Protocol for Cross-Silo Federated Learning with Coding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4HAIYDTS}},
  note         = {Machine review of arXiv:2501.00216}
}
read the original abstract

Federated Learning (FL) is an innovative distributed machine learning paradigm that enables multiple parties to collaboratively train a model without sharing their raw data, thereby preserving data privacy. Communication efficiency concerns arise in cross-silo FL, particularly due to the network heterogeneity and fluctuations associated with geo-distributed silos. Most existing solutions to these problems focus on algorithmic improvements that alter the FL algorithm but sacrificing the training performance. How to address these problems from a network perspective that is decoupled from the FL algorithm remains an open challenge. In this paper, we propose FedCod, a new application layer communication protocol designed for cross-silo FL. FedCod transparently utilizes a coding mechanism to enhance the efficient use of idle bandwidth through client-to-client communication, and dynamically adjusts coding redundancy to mitigate network bottlenecks and fluctuations, thereby improving the communication efficiency and accelerating the training process. In our real-world experiments, FedCod demonstrates a significant reduction in average communication time by up to 62% compared to the baseline, while maintaining FL training performance and optimizing inter-client communication traffic.

Figures

Figures reproduced from arXiv: 2501.00216 by the authors.

Figure 1
Figure 1. Detailed location information and communication bandwidth profiling results: (a) Global topology; (b) North America topology; (c) Profiling results for the global topology; (d) Profiling results for the North America topology. Baseline Network Coding (Download) Network Coding (Upload) Communication Time (s) 0 20 40 Download Upload Waiting [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The communication overhead for baseline and our two adapted network coding communication protocols (one in the download phase and the other in the upload phase). We provide some intuition for this observation. Client￾to-client communication allows sinks to use idle bandwidth for model weight exchange. Additionally, coding redundancy enhances the efficiency of client-to-client communication and increases system toler… view at source ↗
Figure 3
Figure 3. The download phase of FEDCOD (k = 2). The server encodes the global model partitions with random coefficient vectors. Step ❶: The server sends different encoded data blocks to different clients; Step ❷: The clients send the data blocks in the buffer to all neighbors; Step ❸: The client decodes a global model with enough encoded data blocks. cient vector At i = (α t i,1 , αt i,2 , . . . , αt i,k) T , i.e., Mt i = Gt … view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The upload phase of FEDCOD with Coded-AGR algorithm (k = 2). All clients encode the local model updates with the same sequence of coefficient vectors. Step ❶: The clients send the encoded data blocks to the specific neighbors with the predefined mapping; Step ❷: Each c…
Figure 5
Figure 5. Figure 5: The communication time of different communication protocols. Baseline: the basic server-client communication protocol; HierFL: the hierarchical FL communication protocol; D1-NC: apply network coding in the download phase; D2- C: apply our coding strategy in the downloa…
Figure 6
Figure 6. Figure 6: The communication time for each client of different communication protocols. TABLE I: The average server traffic (Unit: MBytes) Topology Protocol Ingress Egress Global Baseline 4003.697 4003.697 D1-NC 4003.697 2342.916 D2-C 4003.697 1334.572 U1-C 7612.005 4003.697 U2-A…
Figure 8
Figure 8. Figure 8: The impact of the number of model partitions k. munication traffic between FEDCOD with static redundancy and with adaptive redundancy. For the global topology, [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 7
Figure 7. Figure 7: The real-world distribution of the communication bandwidth between each pair of silos in the global topology. the upload phase, our coding strategy with wait mode Coded￾AGR (U3-AGR) reduces the upload communication time by 27% and decreases server ingress traffic to ju…
Figure 9
Figure 9. Figure 9: The communication time for different redundancy and faulty links in the global topology. The reason is that large k give the model partitions more opportunity to be aggregated and forwarded to the server by clients with fast transmission speeds. For both the download a…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Benchmark Evaluation of Federated Learning on Multi-organ Images

    cs.CV 2026-07 conditional novelty 6.0 of 10

    MobenFL is the broadest federated medical-imaging benchmark to date, pairing 20 algorithms with 22 multi-organ datasets and adding efficiency plus privacy metrics.

Reference graph

Works this paper leans on

47 extracted references · 44 canonical work pages · cited by 1 Pith paper

  1. [1]

    Federated Optimiza- tion: Distributed Optimization Beyond the Datacenter,

    J. Kone ˇcn`y, B. McMahan, and D. Ramage, “Federated Optimiza- tion: Distributed Optimization Beyond the Datacenter,” arXiv preprint arXiv:1511.03575, 2015

    arXiv 2015

  2. [2]

    Federated Optimization in Heterogeneous Networks,

    T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, and V . Smith, “Federated Optimization in Heterogeneous Networks,” in Proc. MLSys, 2020

    work page 2020

  3. [3]

    Advances and Open Problems in Federated Learning,

    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 , vol. 14, no. 1–2, pp. 1–210, 2021

    work page 2021

  4. [4]

    Throughput-Optimal Topology Design for Cross-Silo Federated Learning,

    O. Marfoq, C. Xu, G. Neglia, and R. Vidal, “Throughput-Optimal Topology Design for Cross-Silo Federated Learning,” in Proc. NeurIPS, 2020

    work page 2020

  5. [5]

    Industry-Scale Orchestrated Federated Learning for Drug Discovery,

    M. Oldenhof, G. ´Acs, B. Pej ´o, A. Schuffenhauer, N. Holway, N. Sturm, A. Dieckmann, O. Fortmeier, E. Boniface, C. Mayer et al. , “Industry-Scale Orchestrated Federated Learning for Drug Discovery,” in Proc. AAAI, 2023

    work page 2023

  6. [6]

    Federated Learning in Distributed Medical Databases: Meta-analysis of Large-Scale Subcortical Brain Data,

    S. Silva, B. A. Gutman, E. Romero, P. M. Thompson, A. Altmann, and M. Lorenzi, “Federated Learning in Distributed Medical Databases: Meta-analysis of Large-Scale Subcortical Brain Data,” in 2019 IEEE 16th international symposium on biomedical imaging (ISBI 2019), 2019

    work page 2019

  7. [7]

    Deep Learning-Based Classification of Mesothelioma Improves Prediction of Patient Outcome,

    P. Courtiol, C. Maussion, M. Moarii, E. Pronier, S. Pilcer, M. Sefta, P. Manceron, S. Toldo, M. Zaslavskiy, N. Le Stang et al. , “Deep Learning-Based Classification of Mesothelioma Improves Prediction of Patient Outcome,” Nature Medicine , vol. 25, no. 10, pp. 1519–1525, 2019

    work page 2019

  8. [8]

    FATE: An Industrial Grade Platform for Collaborative Learning With Data Protection,

    Y . Liu, T. Fan, T. Chen, Q. Xu, and Q. Yang, “FATE: An Industrial Grade Platform for Collaborative Learning With Data Protection,” Journal of Machine Learning Research , 2021

    work page 2021

Show all 47 references
  1. [9]

    Artificial Intelligence Act,

    T. Madiega, “Artificial Intelligence Act,” European Parliament: Euro- pean Parliamentary Research Service , 2021

    2021

  2. [10]

    Federated Learning Priorities Under the European Union Artificial Intelligence Act,

    H. Woisetschl ¨ager, A. Erben, B. Marino, S. Wang, N. D. Lane, R. Mayer, and H.-A. Jacobsen, “Federated Learning Priorities Under the European Union Artificial Intelligence Act,” arXiv preprint arXiv:2402.05968 , 2024

    2024 arXiv

  3. [11]

    FATE-LLM: A Industrial Grade Federated Learning Framework for Large Language Models,

    T. Fan, Y . Kang, G. Ma, W. Chen, W. Wei, L. Fan, and Q. Yang, “FATE-LLM: A Industrial Grade Federated Learning Framework for Large Language Models,” arXiv preprint arXiv:2310.10049 , 2023

    2023 arXiv

  4. [12]

    Towards Building the FederatedGPT: Federated Instruction Tuning,

    J. Zhang, S. Vahidian, M. Kuo, C. Li, R. Zhang, T. Yu, G. Wang, and Y . Chen, “Towards Building the FederatedGPT: Federated Instruction Tuning,” in Proc. ICASSP, 2024

    2024

  5. [13]

    Gaia: Geo-Distributed Machine Learning Approaching LAN Speeds,

    K. Hsieh, A. Harlap, N. Vijaykumar, D. Konomis, G. R. Ganger, P. B. Gibbons, and O. Mutlu, “Gaia: Geo-Distributed Machine Learning Approaching LAN Speeds,” in Proc. NSDI, 2017

    2017

  6. [14]

    Robust and Communication-Efficient Federated Learning From Non-i.i.d. Data,

    F. Sattler, S. Wiedemann, K.-R. M ¨uller, and W. Samek, “Robust and Communication-Efficient Federated Learning From Non-i.i.d. Data,” IEEE Transactions on Neural Networks and Learning Systems , vol. 31, no. 9, pp. 3400–3413, 2019

    2019

  7. [15]

    Heterogeneous Federated Learning: State-of-the-Art and Research Challenges,

    M. Ye, X. Fang, B. Du, P. C. Yuen, and D. Tao, “Heterogeneous Federated Learning: State-of-the-Art and Research Challenges,” ACM Computing Surveys, vol. 56, no. 3, pp. 1–44, 2023

    2023

  8. [16]

    Annulus: A Dual Congestion Control Loop for Datacenter and Wan Traffic Aggregates,

    A. Saeed, V . Gupta, P. Goyal, M. Sharif, R. Pan, M. Ammar, E. Zegura, K. Jang, M. Alizadeh, A. Kabbani et al., “Annulus: A Dual Congestion Control Loop for Datacenter and Wan Traffic Aggregates,” in Proc. SIG- COMM, 2020

    2020

  9. [17]

    Examination of W AN Traffic Characteristics in a Large-Scale Data Center Network,

    Z. Wang, Z. Li, G. Liu, Y . Chen, Q. Wu, and G. Cheng, “Examination of W AN Traffic Characteristics in a Large-Scale Data Center Network,” in Proc. IMC, 2021

    2021

  10. [18]

    FetchSGD: Communication-Efficient Fed- erated Learning With Sketching,

    D. Rothchild, A. Panda, E. Ullah, N. Ivkin, I. Stoica, V . Braverman, J. Gonzalez, and R. Arora, “FetchSGD: Communication-Efficient Fed- erated Learning With Sketching,” in Proc. ICML, 2020

    2020

  11. [19]

    Do- CoFL: Downlink Compression for Cross-Device Federated Learning,

    R. Dorfman, S. Vargaftik, Y . Ben-Itzhak, and K. Y . Levy, “Do- CoFL: Downlink Compression for Cross-Device Federated Learning,” in Proc. ICML, 2023

    2023

  12. [20]

    Local SGD Converges Fast and Communicates Little,

    S. U. Stich, “Local SGD Converges Fast and Communicates Little,” in Proc. ICLR, 2019

    2019

  13. [21]

    Cost-Effective Federated Learning Design,

    B. Luo, X. Li, S. Wang, J. Huang, and L. Tassiulas, “Cost-Effective Federated Learning Design,” in Proc. INFOCOM, 2021

    2021

  14. [22]

    DENSE: Data-Free One-shot Federated Learning,

    J. Zhang, C. Chen, B. Li, L. Lyu, S. Wu, S. Ding, C. Shen, and C. Wu, “DENSE: Data-Free One-shot Federated Learning,” in Proc. NeurIPS, 2022

    2022

  15. [23]

    D2: Decentralized Training Over Decentralized Data,

    H. Tang, X. Lian, M. Yan, C. Zhang, and J. Liu, “ D2: Decentralized Training Over Decentralized Data,” in Proc. ICML, 2018

    2018

  16. [24]

    ClusterFL: A Clustering-based Federated Learning System for Human Activity Recognition,

    X. Ouyang, Z. Xie, J. Zhou, G. Xing, and J. Huang, “ClusterFL: A Clustering-based Federated Learning System for Human Activity Recognition,” ACM Transactions on Sensor Networks , vol. 19, no. 1, pp. 1–32, 2022

    2022

  17. [25]

    SCAFFOLD: Stochastic Controlled Averaging for Federated Learning,

    S. P. Karimireddy, S. Kale, M. Mohri, S. Reddi, S. Stich, and A. T. Suresh, “SCAFFOLD: Stochastic Controlled Averaging for Federated Learning,” in Proc. ICML, 2020

    2020

  18. [26]

    Hierarchical Federated Learning With Quantization: Convergence Analysis and System Design,

    L. Liu, J. Zhang, S. Song, and K. B. Letaief, “Hierarchical Federated Learning With Quantization: Convergence Analysis and System Design,” IEEE Transactions on Wireless Communications , vol. 22, no. 1, pp. 2– 18, 2022

    2022

  19. [27]

    On the Convergence of Hier- archical Federated Learning with Gradient Quantization and Imperfect Transmission,

    H. Sun, H. Tian, W. Ni, and J. Zheng, “On the Convergence of Hier- archical Federated Learning with Gradient Quantization and Imperfect Transmission,” in Proc. ICASSP, 2024

    2024

  20. [28]

    Which Mode Is Better for Federated Learning? Centralized or Decentralized,

    Y . Sun, L. Shen, and D. Tao, “Which Mode Is Better for Federated Learning? Centralized or Decentralized,” arXiv preprint arXiv:2310.03461, 2023

    2023 arXiv

  21. [29]

    Hierarchi- cal Federated Learning Across Heterogeneous Cellular Networks,

    M. S. H. Abad, E. Ozfatura, D. Gunduz, and O. Ercetin, “Hierarchi- cal Federated Learning Across Heterogeneous Cellular Networks,” in Proc. ICASSP, 2020

    2020

  22. [30]

    Resource- Efficient Federated Learning With Hierarchical Aggregation in Edge Computing,

    Z. Wang, H. Xu, J. Liu, H. Huang, C. Qiao, and Y . Zhao, “Resource- Efficient Federated Learning With Hierarchical Aggregation in Edge Computing,” in Proc. INFOCOM, 2021

    2021

  23. [31]

    HierFedML: Aggregator Placement and UE Assignment for Hierarchical Federated Learning in Mobile Edge Computing,

    Z. Xu, D. Zhao, W. Liang, O. F. Rana, P. Zhou, M. Li, W. Xu, H. Li, and Q. Xia, “HierFedML: Aggregator Placement and UE Assignment for Hierarchical Federated Learning in Mobile Edge Computing,” IEEE Transactions on Parallel and Distributed Systems , vol. 34, no. 1, pp. 328–345, 2022

    2022

  24. [32]

    Communication-Efficient Learning of Deep Networks From Decentral- ized Data,

    B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, “Communication-Efficient Learning of Deep Networks From Decentral- ized Data,” in Proc. Artificial Intelligence and Statistics , 2017

    2017

  25. [33]

    Revisiting Weighted Aggregation in Federated Learning With Neural Networks,

    Z. Li, T. Lin, X. Shang, and C. Wu, “Revisiting Weighted Aggregation in Federated Learning With Neural Networks,” in Proc. ICML, 2023

    2023

  26. [34]

    Improving Federated Learning With Quality-Aware User Incentive and Auto-Weighted Model Aggregation,

    Y . Deng, F. Lyu, J. Ren, Y .-C. Chen, P. Yang, Y . Zhou, and Y . Zhang, “Improving Federated Learning With Quality-Aware User Incentive and Auto-Weighted Model Aggregation,”IEEE Transactions on Parallel and Distributed Systems, vol. 33, no. 12, pp. 4515–4529, 2022

    2022

  27. [35]

    Echelon: Peer-to-Peer Network Diagnosis With Network Coding,

    C. Wu and B. Li, “Echelon: Peer-to-Peer Network Diagnosis With Network Coding,” in Proc. IWQoS, 2006

    2006

  28. [36]

    Lava: A Reality Check of Network Coding in Peer-to-Peer Live Streaming,

    M. Wang and B. Li, “Lava: A Reality Check of Network Coding in Peer-to-Peer Live Streaming,” in Proc. INFOCOM, 2007

    2007

  29. [37]

    On Large-Scale Peer-to-Peer Streaming Systems With Network Coding,

    C. Feng and B. Li, “On Large-Scale Peer-to-Peer Streaming Systems With Network Coding,” in Proc. ACM MM , 2008

    2008

  30. [38]

    Erasure Coding in Windows Azure Storage,

    C. Huang, H. Simitci, Y . Xu, A. Ogus, B. Calder, P. Gopalan, J. Li, and S. Yekhanin, “Erasure Coding in Windows Azure Storage,” in Proc. USENIX ATC, 2012, pp. 15–26

    2012

  31. [39]

    A Random Linear Network Coding Approach to Multicast,

    T. Ho, M. M ´edard, R. Koetter, D. R. Karger, M. Effros, J. Shi, and B. Leong, “A Random Linear Network Coding Approach to Multicast,” IEEE Transactions on Information Theory , vol. 52, no. 10, pp. 4413– 4430, 2006

    2006

  32. [40]

    How Practical Is Network Coding?

    M. Wang and B. Li, “How Practical Is Network Coding?” in Proc. IWQoS, 2006

    2006

  33. [41]

    Random Network Coding in Peer-to-Peer Networks: From Theory to Practice,

    B. Li and D. Niu, “Random Network Coding in Peer-to-Peer Networks: From Theory to Practice,” Proceedings of the IEEE , vol. 99, no. 3, pp. 513–523, 2011

    2011

  34. [42]

    A. L. Cauchy, Exercices d’analyse et de physique mathematique: 1 . Bachelier, imprimeur-libraire, 1840, vol. 1

  35. [43]

    Optimizing Cauchy Reed-Solomon Codes for Fault-tolerant Network Storage Applications,

    J. S. Plank and L. Xu, “Optimizing Cauchy Reed-Solomon Codes for Fault-tolerant Network Storage Applications,” in Proc. NCA, 2006

    2006

  36. [44]

    FedLab: A Flex- ible Federated Learning Framework,

    D. Zeng, S. Liang, X. Hu, H. Wang, and Z. Xu, “FedLab: A Flex- ible Federated Learning Framework,” Journal of Machine Learning Research, vol. 24, no. 100, pp. 1–7, 2023

    2023

  37. [45]

    Network Information Flow,

    R. Ahlswede, N. Cai, S.-Y . Li, and R. W. Yeung, “Network Information Flow,” IEEE Transactions on Information Theory , vol. 46, no. 4, pp. 1204–1216, 2000

    2000

  38. [46]

    Algebraic Gossip: A Network Coding Approach to Optimal Multiple Rumor Mongering,

    S. Deb, M. M ´edard, and C. Choute, “Algebraic Gossip: A Network Coding Approach to Optimal Multiple Rumor Mongering,” IEEE Trans- actions on Information Theory , vol. 52, no. 6, pp. 2486–2507, 2006

    2006

  39. [47]

    Network Coding for Federated Learning Systems,

    L. Kong, H. Tao, J. Wang, Z. Huang, and J. Xiao, “Network Coding for Federated Learning Systems,” in Proc. ICONIP, 2020

    2020

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.