Quantized Rank Reduction: A Communications-Efficient Federated Learning Scheme for Network-Critical Applications

Constantine Kotropoulos; Dimitrios Kritsiolis

REVIEW 3 major objections 6 minor 26 references

Quantized Rank Reduction: A Communications-Efficient Federated Learning Scheme for Network-Critical Applications

T0 review · 3 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A federated learning scheme compresses each client's gradient update by truncated SVD or Tucker decomposition and quantization, reporting near-baseline accuracy at a small fraction of the bandwidth.

desk verdict A straightforward low-rank-plus-quantization combination with honest bit accounting, but the 'minimal impact' accuracy claim is contradicted by the paper's own CIFAR-10 result and the experiments are too thin. read the letter →

arxiv 2507.11183 v1 pith:X3N55KN2 submitted 2025-07-15 cs.LG

Dimitrios Kritsiolis , Constantine Kotropoulos This is my paper · ORCID

classification cs.LG

keywords federatedlearningcommunicationefficiencylow-rankapproximationtruncatedSVDTuckerdecompositiongradientquantizationnetwork-criticalapplications

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proposes Quantized Rank Reduction (QRR), a scheme for communication-efficient federated learning. It compresses the gradient of each layer before upload: a fully connected layer's gradient matrix is truncated by SVD, a convolutional layer's gradient tensor by Tucker decomposition, and all resulting factors are quantized with a lazily aggregated quantizer. The authors claim that QRR achieves accuracy close to standard federated averaging and to SLAQ while transmitting 3.34-9.43% of SGD's bits and 13.52-44.05% of SLAQ's bits across MNIST and CIFAR-10 experiments. The trade-off is slower convergence per training iteration, but faster progress per transmitted bit, which matters for network-critical settings.

What carries the argument

The load-bearing mechanism is the pair of operators $(C, Q)$: compression $C$ maps the gradient matrix or tensor to a truncated SVD or Tucker decomposition controlled by one fraction $p$, and quantization $Q$ maps each resulting factor onto a $\beta$-bit grid whose radius is the $\ell^\infty$ distance from the previous quantized factor. The central identity is the reconstruction on the server, $\hat{g} = Q(U)Q(\Sigma)Q(V)^\top$ for SVD and the equivalent mode-product formula for Tucker, so the argument reduces to counting transmitted bits: only $\nu(D_{\mathrm{out}}+D_{\mathrm{in}}+1)$ scalars instead of $D_{\mathrm{out}}D_{\mathrm{in}}$, and the analogous count for Tucker. The parameter $p<0.5$ balances reconstruction error against bit savings.

What would settle it

Compute the cumulative energy $\sum_{j=1}^{\nu}\sigma_j^2 / \sum_{j=1}^{r}\sigma_j^2$ of actual gradients from a model such as a transformer or ResNet on ImageNet for $\nu = \lceil 0.1\cdot r\rceil$; if the ratio is well below the value needed to keep training stable, then QRR's premise fails for that setting. Alternatively, run QRR with $p=0.1$ on such a model and check whether the accuracy gap to SGD exceeds roughly the 1-9 points seen on MNIST and CIFAR-10.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that the communication bottleneck of federated learning can be attacked by treating each client's gradient update as a low-rank object. For fully connected layers the update matrix is factored as $A \approx U_\nu \Sigma_\nu V_\nu^\top$ retaining the $\nu = \lceil p\cdot \min(D_{\mathrm{out}}, D_{\mathrm{in}})\rceil$ largest singular values, and for convolutional layers the update tensor is factored by Tucker decomposition with per-mode ranks $r_i = \lceil p\cdot I_i\rceil$. The factor matrices and core tensor are then quantized with a $\beta$-bit lazily aggregated quantizer, transmitted, and reconstructed on the server. The reported result is that QRR keeps accuracy within 1-3 points on MNIST and within about 9 points on CIFAR-10 of SGD or SLAQ while transmitting 3.34-9.43% of the bits of SGD and 13.52-44.05% of the bits of SLAQ.

Load-bearing premise

The whole scheme rests on the assumption that neural network gradient matrices and tensors are low-rank enough that keeping a fraction $p<0.5$ of the singular values still carries most of the training signal; if real gradients have a flat singular-value spectrum, the compressed updates lose too much information and the bandwidth-accuracy trade-off collapses.

Editorial extensions

If this is right

With QRR, a federated client can cut its per-iteration upload by one to two orders of magnitude compared with sending raw gradients, at a cost of 1-9 accuracy points depending on dataset and $p$.
On the MNIST experiments, QRR reaches a given accuracy faster than SGD and SLAQ when progress is measured per transmitted bit, meaning bandwidth, not iterations, can be the scarce resource.
On CIFAR-10, the 8-9 point gap suggests QRR is suitable for getting a deployable model quickly on constrained links rather than for final accuracy.
The measured client-side overhead is modest: about 1.2 times the memory and 3.8 times the computation time of SGD, versus 13 times the memory for SLAQ.

Reading between the lines

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

A natural extension is an adaptive $p$ chosen per layer, per client, or per round, since the paper only tests fixed or evenly spaced client-level values; an adaptive scheme would likely improve the accuracy-bandwidth frontier.
If the low-rank assumption weakens for models like large transformers, QRR's accuracy loss is expected to grow; a useful comparison would be the singular-value spectra of attention and MLP gradients to identify which layers tolerate aggressive truncation.
The paper does not give a convergence bound for the compressed-quantized update, so a formal analysis of when the $p$ parameter preserves convergence is the natural next step.

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.

Desk Editor's Note

read the letter

You should know this paper is a clean, careful combination of two existing techniques: low-rank gradient compression (truncated SVD / Tucker) and LAQ-style quantization. The authors don't claim a new factorization or new quantizer; the novelty is in the integration and in the measured bit-accuracy tradeoffs. The bit accounting in Tables I-III checks out, and the reported savings (down to a few percent of SGD's bits) are real for the setups tested.

The strongest part is that the method is honestly presented. The equations are consistent, the server-side reconstruction from quantized factors is coherent, and the paper reports client-side compute/memory overhead rather than hiding it. The code is linked, though without a commit hash.

The soft spots are real. Section II.A asserts gradient matrices are generally low-rank, citing [14], and supports it with one unlabeled singular-value plot for a single fully-connected layer. No retained-energy measurement (Eq. 7) or Tucker analogue is reported for any layer, iteration, or dataset. That matters because CIFAR-10 is exactly the test of that premise, and it fails: an 8-9 point accuracy drop versus SGD/SLAQ. The abstract's promise of 'minimal impact' is not defensible for that result, and the conclusion's 'slightly lower accuracy' is too broad.

The experimental section is thinner than it should be: single runs, no seeds, no error bars. With 10 clients and a batch size of 512, one more seed would at least tell us whether the 1-3 point gap on MNIST is stable. The comparison set is narrow (SGD and SLAQ only); a rank-compression-only ablation would isolate the contribution of the quantizer. The claims about network-critical applications and 'quickly reaching a deployable model state' are speculative.

Who is this for? People building communication-efficient FL systems who want a straightforward baseline that combines low-rank and quantization. It is not a breakthrough, but it is a sensible, correctly-accounted engineering contribution.

For peer review: I would send it out. A careful referee will need to insist on repeated runs and a rank-energy check, but the method deserves referee time rather than desk rejection.

Referee Report

3 major / 6 minor

Summary. The paper proposes Quantized Rank Reduction (QRR), a federated-learning compression scheme that combines low-rank matrix/tensor factorization (truncated SVD for fully-connected layers, Tucker decomposition for convolutional layers) with the LAQ quantized-gradient framework. Each client compresses its gradient update, quantizes the factor components with β bits, and transmits the quantized factors; the server reconstructs an approximate gradient and performs distributed gradient descent. The authors derive bit-count comparisons from the factor sizes, report experiments on an MLP and a CNN for MNIST and a VGG-like CNN for CIFAR-10, and compare QRR against SGD and SLAQ. The headline claim is that QRR transmits a small fraction of the bits of SGD or SLAQ while maintaining, on MNIST, accuracy within 1–3 percentage points of the baselines; on CIFAR-10 the accuracy drop is about 9 percentage points.

Significance. If the underlying low-rank premise holds broadly, the bit-accounting part of the paper is sound and directly useful: the reported bit counts follow from the rank-retention formula and the quantization width, and the code is made available, which aids reproducibility. The combination of rank reduction with lazily aggregated quantization is a reasonable design, and the comparison against SLAQ is appropriate. However, the paper's evidence for the load-bearing premise is thin, and its own CIFAR-10 result contradicts the abstract's 'minimal impact on accuracy' wording. As it stands, the paper establishes a bandwidth-accuracy tradeoff on MNIST-like tasks, but not a general method with minimal accuracy impact. The central idea may be salvageable with additional measurements and more careful claims, so a major revision is warranted rather than rejection.

major comments (3)

[Section II.A, Eq. (7), Figure 1] The paper's central premise is that neural-network gradient matrices and tensors 'are generally low-rank and have a few dominant singular values,' but the only supporting evidence is one unlabeled singular-value plot of a single fully-connected layer gradient and a citation to [14]. No quantitative measurement of the retained Frobenius energy, as defined in Eq. (7), or of a Tucker analogue is reported for any layer, iteration, or dataset. This is load-bearing because if the gradients are not sufficiently low-rank, truncating to p<0.5 discards signal and the bandwidth-accuracy tradeoff collapses. Please add measurements of the form ||A−A_ν||_F^2 / ||A||_F^2 for the SVD case and the corresponding Tucker reconstruction error for each layer type and each p value, across iterations, for all three experiments.
[Table III and Section IV] The conclusion states that QRR has 'slightly lower accuracy' than Federated Averaging or SLAQ, and the abstract promises 'minimal impact on the model's accuracy.' Table III does not support this: on CIFAR-10 QRR reaches 47.57% accuracy versus 56.72% for SGD and 55.73% for SLAQ, a drop of about 8–9 percentage points, with a correspondingly worse loss (1.441 versus 1.213 and 1.242). The text in Section III.B acknowledges the reduced accuracy on the more complex dataset, but the global claims in the abstract and conclusions are too broad. Please either temper the claims to reflect the dataset-dependent nature of the accuracy loss or add a mechanism, such as energy-based per-layer rank selection, that keeps the accuracy drop small on harder tasks.
[Section III.B, Tables I–III] All experiments appear to be single runs: no random seeds, no repeated trials, and no variance estimates are reported. With 10 clients and stochastic batches, the 1–2 percentage point differences on MNIST may be within run-to-run noise, and the claimed 'faster convergence in terms of bits' could depend on initialization. Please report mean and standard deviation over at least several seeds, or clearly state that the tables are illustrative single runs and avoid over-interpreting small accuracy gaps.

minor comments (6)

[Section II.A, Figure 1] Figure 1 lacks axis labels and a description of which layer, dataset, and iteration produced the gradient; please add these details and state whether the singular values are normalized.
[Section II.A, Eq. (7)] The notation 'rX' in the summation is a typo; it should read a summation over j from ν+1 to r, i.e., \sum_{j=\nu+1}^{r} \sigma_j^2.
[Section III.A, Eq. (19)] The relationship between the QRR_c operator defined in Eq. (19) and the later description of compression followed by quantization is not fully explicit; a short pseudocode or algorithm box would clarify the order of operations and the client-side state that must be stored.
[Section III.B] The SLAQ parameters 'D = 10, ξ1,...,ξD = 1/D' are introduced without defining D and ξ; please define these symbols or cite the specific definition from [22].
[Table III] For the CIFAR-10 experiment, the paper says that each client is assigned a p value evenly spaced in [0.1, 0.3], but the table does not list those p values or an average retention ratio; please report them so the bit count can be checked.
[Figures 2–4] The vertical axis label in Figure 4 is simply 'Gradient'; for consistency with Figures 2 and 3 it should read 'Gradient ℓ2 norm'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: QRR's accuracy and bit-reduction claims are empirical results predicated on an external low-rank-gradient assumption, not on fitting parameters to the headline quantities.

full rationale

The central claim—that QRR transmits a small fraction of SGD/SLAQ bits at a modest accuracy cost—is supported by the experimental tables (Tables I–III), which report measured loss, accuracy, and bit counts computed from the explicit SVD/Tucker and quantization formulas. The compression parameter p and quantization bits β are fixed before the runs, and no fitted parameter is later renamed as a prediction; Eq. (19) is a definition of the scheme, not a derivation of the experimental outcomes. The only load-bearing premise, that gradient matrices/tensors are low-rank, is imported from an external reference [14] and a supporting Figure 1; it is a data-dependent assumption and a correctness/generalization risk (especially given the CIFAR-10 9-point accuracy drop), not a circular step. The paper contains no load-bearing self-citation chain: the authors' own repository link [12] is incidental, and the cited low-rank and quantization results [14,22] are independent external work. Because the empirical accuracy-versus-bits comparison is testable and is not constructed by fitting p or β to the reported accuracies, no reduction of the claimed result to its own inputs is present.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The reported tradeoff depends mainly on the low-rank-gradient assumption and on standard SVD, Tucker, and LAQ error bounds. The parameter p and bit width beta are engineering choices scanned by the authors, not fitted to the accuracy target. No new entities are introduced.

free parameters (2)

rank retention ratio p = 0.1, 0.2, 0.3; per-client values in [0.1, 0.3] for CIFAR-10
Controls the SVD/Tucker truncation; the reported bit reductions and accuracy losses depend directly on this choice. It is scanned by the authors, not fitted to a target.
quantization bit width beta = 8 bits
Sets the number of bits per quantized element and therefore the quantization error bound. Chosen as a fixed experimental setting.

assumptions (4)

domain assumption Gradient matrices and tensors of neural network layers are approximately low-rank, with a few dominant singular values.
Section II.A and Figure 1; this justifies truncation and controls the bit savings.
standard math Eckart-Young theorem: the rank-nu truncated SVD minimizes Frobenius error, with error equal to the sum of squared truncated singular values.
Equation (7) in Section II.A.
standard math Tucker decomposition represents a 4-mode tensor with a small core and factor matrices, with reconstruction given by the mode-n product.
Equations (9)-(11) in Section II.A are taken as given.
standard math LAQ quantization with grid radius R and step 2*tau*R has sup-norm error at most tau*R.
Equation (18) in Section II.B follows from the definition of the grid.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Quantized Rank Reduction: A Communications-Efficient Federated Learning Scheme for Network-Critical Applications." pith.science (2026). https://pith.science/paper/X3N55KN2

@misc{pith2026250711183,
  author       = {Pith},
  title        = {Pith review of: Quantized Rank Reduction: A Communications-Efficient Federated Learning Scheme for Network-Critical Applications},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/X3N55KN2}},
  note         = {Machine review of arXiv:2507.11183}
}

read the original abstract

Federated learning is a machine learning approach that enables multiple devices (i.e., agents) to train a shared model cooperatively without exchanging raw data. This technique keeps data localized on user devices, ensuring privacy and security, while each agent trains the model on their own data and only shares model updates. The communication overhead is a significant challenge due to the frequent exchange of model updates between the agents and the central server. In this paper, we propose a communication-efficient federated learning scheme that utilizes low-rank approximation of neural network gradients and quantization to significantly reduce the network load of the decentralized learning process with minimal impact on the model's accuracy.

Figures

Figures reproduced from arXiv: 2507.11183 by the authors.

**Figure 2.** Loss, gradient ℓ2 norm, and accuracy plotted against the number of iterations and bits for the MLP network and the MNIST dataset. max pooling, and dropout layers, with the number of filters increasing from 32 to 64 and then to 128. We used different values of p to demonstrate that p can be chosen based on the client’s connection speed and the amount of data transmitted from that client. Evenly spaced values in [0.1,… view at source ↗

**Figure 4.** Loss, gradient ℓ2 norm, and accuracy plotted against the number of iterations and bits for the VGG-like CNN and the CIFAR-10 dataset. a fraction of the bits required by the other methods. It converges more slowly with the number of iterations, but faster when considering the number of bits transmitted. There is an added computational and memory overhead on both the client and server sides. However, this scheme can p… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 21 canonical work pages

[14]

Generalization guarantees for neural networks via harnessing the low-rank structure of the Jacobian,

S. Oymak, Z. Fabian, M. Li, and M. Soltanolkotabi, “Generalization guarantees for neural networks via harnessing the low-rank structure of the Jacobian,” 2019. [Online]. Available: https://arxiv.org/abs/1906. 05392

work page 2019
[1]

Federated learning: Strategies for improving communication efficiency,

J. Kone ˇcn`y, H. B. McMahan, F. X. Yu, P. Richt ´arik, A. T. Suresh, and D. Bacon, “Federated learning: Strategies for improving communication efficiency,” 2017. [Online]. Available: https://arxiv.org/abs/1610.05492

arXiv 2017
[2]

A survey on federated learning,

C. Zhang et al. , “A survey on federated learning,” Knowledge-Based Systems, vol. 216, p. 106775, 2021

work page 2021
[3]

Federated learning in mobile edge networks: A com- prehensive survey,

W. Lim et al. , “Federated learning in mobile edge networks: A com- prehensive survey,”IEEE Communications Surveys & Tutorials, vol. 22, no. 3, pp. 2031–2063, 2020

work page 2020
[4]

Limitations and future aspects of communication costs in federated learning: A survey,

M. Asad et al., “Limitations and future aspects of communication costs in federated learning: A survey,” Sensors, vol. 23, no. 17, p. 7358, 2023

work page 2023
[5]

Advances and open problems in federated learning,

P. Kairouz 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
[6]

Communication efficient distributed machine learning with the parameter server,

M. Li, D. G. Andersen, A. Smola, and K. Yu, “Communication efficient distributed machine learning with the parameter server,” in Advances in Neural Information Processing Systems , 2014, vol. 27, pp. 19—-27

work page 2014
[7]

Communication complexity of distributed convex learning and optimization,

Y . Arjevani and O. Shamir, “Communication complexity of distributed convex learning and optimization,” in Advances in Neural Information Processing Systems, 2015, vol. 28, pp. 1756––1764

work page 2015

Show all 26 references

[8]

Communication-efficient learning of deep networks from decentralized data,

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

2017
[9]

Efficient neural network compression,

H. Kim, M. U. K. Khan, and C.-M. Kyung, “Efficient neural network compression,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2019, pp. 12 569–12 577

2019
[10]

Marvel: Towards efficient federated learning on IoT devices,

L. Liu and X. Xu, “Marvel: Towards efficient federated learning on IoT devices,” Computer Networks, vol. 245, p. 110375, 2024

2024
[11]

Deep neural network compression by Tucker decomposition with nonlinear response,

Y . Liu and M. K. Ng, “Deep neural network compression by Tucker decomposition with nonlinear response,” Knowledge-Based Systems, vol. 241, p. 108171, 2022

2022
[12]

Quantized rank reduction: A communications-efficient federated learning scheme for network-critical applications,

“Quantized rank reduction: A communications-efficient federated learning scheme for network-critical applications,” [retrieved: May 22, 2025]. [Online]. Available: https://github.com/Kritsos/QRR-code

2025
[13]

Computing neural network gradients,

K. Clark, “Computing neural network gradients,” Stanford University, August 2018, Notes

2018
[15]

Some mathematical notes on three-mode factor analysis,

L. R. Tucker, “Some mathematical notes on three-mode factor analysis,” Psychometrika, vol. 31, no. 3, pp. 279–311, 1966

1966
[16]

Compression and interpretability of deep neural networks via Tucker tensor layer: From first principles to tensor valued back-propagation,

G. G. Calvi, A. Moniri, M. Mahfouz, Q. Zhao, and D. P. Mandic, “Compression and interpretability of deep neural networks via Tucker tensor layer: From first principles to tensor valued back-propagation,”
[17]

Tensor-factorized neural networks,

J.-T. Chien and Y .-T. Bao, “Tensor-factorized neural networks,” IEEE Transactions on Neural Networks and Learning Systems , vol. 29, no. 5, pp. 1998–2011, 2017

1998
[18]

De Lathauwer, Signal Processing Based on Multilinear Algebra

L. De Lathauwer, Signal Processing Based on Multilinear Algebra . Katholieke Universiteit Leuven, 1997

1997
[19]

QSGD: Communication-efficient SGD via gradient quantization and encoding,

D. Alistarh, D. Grubic, J. Li, R. Tomioka, and M. V ojnovic, “QSGD: Communication-efficient SGD via gradient quantization and encoding,” in Advances in Neural Information Processing Systems , 2017, vol. 30, pp. 1707––1718

2017
[20]

Distributed learning with compressed gradient differences,

K. Mishchenko, E. Gorbunov, M. Tak ´aˇc, and P. Richt ´arik, “Distributed learning with compressed gradient differences,” 2023. [Online]. Available: https://arxiv.org/abs/1901.09269

2023 arXiv
[21]

Neu- ral network quantization in federated learning at the edge,

N. Tonellotto, A. Gotta, F. M. Nardini, D. Gadler, and F. Silvestri, “Neu- ral network quantization in federated learning at the edge,” Information Sciences, vol. 575, pp. 417–436, 2021

2021
[22]

Lazily aggregated quantized gradient innovation for communication-efficient federated learning,

J. Sun, T. Chen, G. B. Giannakis, Q. Yang, and Z. Yang, “Lazily aggregated quantized gradient innovation for communication-efficient federated learning,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 4, pp. 2031–2044, 2020

2020
[23]

The MNIST database of handwritten digit images for machine learning research,

L. Deng, “The MNIST database of handwritten digit images for machine learning research,” IEEE Signal Processing Magazine, vol. 29, no. 6, pp. 141–142, 2012

2012
[24]

Learning multiple layers of features from tiny images,

A. Krizhevsky, “Learning multiple layers of features from tiny images,” University of Toronto, Tech. Rep., 2009. [Online]. Available: https://www.cs.toronto.edu/∼kriz/learning-features-2009-TR.pdf

2009
[25]

Very deep convolutional networks for large-scale image recognition,

K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” 2015. [Online]. Available: https: //arxiv.org/abs/1409.1556

2015 arXiv
[2019]

Available: https://arxiv.org/abs/1903.06133

[Online]. Available: https://arxiv.org/abs/1903.06133

1903 arXiv

Pith tools

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

[1] [14]

Generalization guarantees for neural networks via harnessing the low-rank structure of the Jacobian,

S. Oymak, Z. Fabian, M. Li, and M. Soltanolkotabi, “Generalization guarantees for neural networks via harnessing the low-rank structure of the Jacobian,” 2019. [Online]. Available: https://arxiv.org/abs/1906. 05392

work page 2019

[2] [1]

Federated learning: Strategies for improving communication efficiency,

J. Kone ˇcn`y, H. B. McMahan, F. X. Yu, P. Richt ´arik, A. T. Suresh, and D. Bacon, “Federated learning: Strategies for improving communication efficiency,” 2017. [Online]. Available: https://arxiv.org/abs/1610.05492

arXiv 2017

[3] [2]

A survey on federated learning,

C. Zhang et al. , “A survey on federated learning,” Knowledge-Based Systems, vol. 216, p. 106775, 2021

work page 2021

[4] [3]

Federated learning in mobile edge networks: A com- prehensive survey,

W. Lim et al. , “Federated learning in mobile edge networks: A com- prehensive survey,”IEEE Communications Surveys & Tutorials, vol. 22, no. 3, pp. 2031–2063, 2020

work page 2020

[5] [4]

Limitations and future aspects of communication costs in federated learning: A survey,

M. Asad et al., “Limitations and future aspects of communication costs in federated learning: A survey,” Sensors, vol. 23, no. 17, p. 7358, 2023

work page 2023

[6] [5]

Advances and open problems in federated learning,

P. Kairouz 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

[7] [6]

Communication efficient distributed machine learning with the parameter server,

M. Li, D. G. Andersen, A. Smola, and K. Yu, “Communication efficient distributed machine learning with the parameter server,” in Advances in Neural Information Processing Systems , 2014, vol. 27, pp. 19—-27

work page 2014

[8] [7]

Communication complexity of distributed convex learning and optimization,

Y . Arjevani and O. Shamir, “Communication complexity of distributed convex learning and optimization,” in Advances in Neural Information Processing Systems, 2015, vol. 28, pp. 1756––1764

work page 2015