Pith. sign in

REVIEW 4 major objections 4 minor 68 references

FLMarket: Enabling Privacy-preserved Pre-training Data Pricing for Federated Learning

T0 review · 4 major / 4 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 FLMarket prices federated clients before training via a truthful auction over secret-shared statistics, and claims over 10% higher downstream accuracy than prior pre-training selection.

desk verdict A genuine integration of auction pricing with privacy-preserving distribution aggregation for FL, but the headline accuracy gain is inflated by a CIFAR-10 tuning loop and missing error bars. read the letter →

arxiv 2411.11713 v1 pith:YZQ3J4QE submitted 2024-11-18 cs.LG cs.DC

Zhenyu Wen , Wanglei Feng , Di Wu , Haozhen Hu , Chang Xu , Bin Qian , Zhen Hong , Cong Wang
show 1 more author
Shouling Ji
This is my paper · ORCID
classification cs.LGcs.DC
keywords federatedlearningdatapricingpre-trainingclientselectionauctionmechanismprivacy-preservingaggregationsecretsharingbudget-feasible
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

FLMarket addresses a gap in federated learning: pricing clients' data before training starts, when no model feedback exists. The paper claims that a two-stage auction—first scoring each client by data volume and class scarcity, then selecting winners under a budget with critical-value payments—picks clients whose data produces over 10% higher downstream accuracy than state-of-the-art pre-training selection, and over 2% higher accuracy with about 3x less runtime than an in-training selector that re-evaluates clients each round. To avoid leaking the class distributions that feed the scores, FLMarket adds a pairwise noise-masking protocol (PASS) whose random terms cancel when all masked distributions are summed. A sympathetic reader would care because pre-training pricing gives clients a credible reward before they commit resources, which the paper's own survey suggests is a practical obstacle to participation in FL data markets.

What carries the argument

The carrying object is the two-stage, budget-constrained pricing mechanism coupled with the PASS privacy protocol. In stage one, the score function $u_e = \sum_c \theta_c \phi(n^c_e)$ with $\theta_c = 1 - n^s_c/N_s$ and $\phi(x) = \sum_{t=1}^{x} -\ln(\min(t/\alpha, 1))$ converts each client's class-count vector into a single value that rewards data volume with diminishing returns and rewards globally scarce classes. In stage two, Algorithm 1 sorts clients by score per bid, admits winners under the budget condition $b_e \le \frac{R}{2}\cdot\frac{u_e}{U(S_k\cup\{e\})}$, and pays each winner the critical value at which it would lose the auction, which makes the mechanism truthful, individually rational, and budget-feasible. PASS supplies the privacy layer: each client adds signed pseudorandom outputs derived from pairwise Diffie-Hellman seeds to its local distribution, and the sum of all masked distributions cancels the noise to recover $N_s$ without revealing any individual $N_e$.

What would settle it

Make one client abort after Diffie-Hellman keys are exchanged but before it sends its masked distribution Y_e; Equation (19) can no longer cancel the pairwise PRG noises, so the server cannot recover the global distribution N_s and the pre-training price cannot be computed, directly testing the all-clients-complete premise.

Watch

Extended reading notes

Core claim

This paper claims that data for federated learning can be priced before training by viewing each client's class distribution through the score function $u_e = \sum_c \theta_c \phi(n^c_e)$, where $\phi$ is a diminishing-returns curve and $\theta_c$ rewards classes that are globally scarce. The server never sees raw local distributions; clients mask them with pairwise Diffie-Hellman-derived random vectors, and the server sums the masks to recover only the global distribution $N_s$. A budget-feasible auction then ranks clients by score per bid, selects winners until the budget binds, and pays each winner its critical value, giving truthful, individually rational, budget-feasible prices. The paper asserts that this selection transfers to FL accuracy: across 57 test configurations on CIFAR-10, CINIC-10, and DEAP, FLMarket records the highest accuracy in 44 distributions and beats the four pre-training baselines by an average of 10.18% accuracy, while matching or exceeding the in-training baseline S-FedAvg with a 3.16x per-round speedup.

Load-bearing premise

The whole pricing pipeline assumes every enrolled client completes the PASS masking step and submits its masked distribution; if any client drops out after key agreement, the pairwise noises do not cancel and the server cannot recover the global class distribution needed to set prices.

Editorial extensions

If this is right

  • Under the paper's evaluation, FLMarket's pre-training selection wins 44 of 57 tested configurations and raises average accuracy by 10.18% over the four pre-training baselines, with the largest gains on the most imbalanced distributions.
  • Because scoring happens once before training, FLMarket adds no per-round selection overhead; the paper reports reaching 0.6 accuracy on CIFAR-10 in about 130 rounds versus 170-550 rounds for the baselines.
  • Against the in-training S-FedAvg baseline, FLMarket achieves comparable or better final accuracy on most tested distributions with a 3.16x average per-round runtime speedup.
  • The payment equals the critical value, so no client can profit by bidding above its true cost, and total payments stay within the buyer's budget R.
  • When the client pool is scaled to 100 clients, FLMarket still shows positive average accuracy gains over the baselines, though the margin shrinks relative to the 20-client setting.

Reading between the lines

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

  • The scarcity-weighting premise could be tested outside the paper's datasets: on any benchmark where scarce classes carry noisy or mislabeled examples, $\theta_c$ would inflate scores for clients the training would be better off ignoring, so validation-based calibration of $\theta_c$ would be a natural extension.
  • PASS has no dropout path as written; in cross-device FL where clients go offline, the protocol would need threshold secret sharing or server-side handling of missing masks before it can be deployed unchanged.
  • The reported 10% figure is a property of client selection, not of the training algorithm; because the paper notes FLMarket can be stacked with in-training methods, the gains should be read as the value of a better starting client set.
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 / 4 minor

Summary. The paper proposes FLMarket, a framework for pre-training data pricing in federated learning. It combines a two-stage auction mechanism (winner selection and payment determination) with a privacy-preserving protocol (PASS) that aggregates clients' class distributions without revealing them. The score function values each client by a combination of data volume and global class scarcity, using a diminishing-return function. The authors prove truthfulness, individual rationality, and budget feasibility of the auction, and evaluate client selection by training on CIFAR-10, CINIC-10, and DEAP, reporting accuracy and runtime comparisons against four pre-training baselines and one in-training baseline.

Significance. The paper addresses a genuinely open problem—pricing data before training in FL—and the proposed mechanism is coherent: the auction is of independent interest, the score function is interpretable, and the evaluation covers a large configuration space (three datasets, multiple selection ratios and distributions). The formal statements in Section 3.3 are plausible, the PASS running example in Appendix F is helpful, and the appendix contains substantial supporting material, including a survey and a proof appendix. However, the headline performance claim is not currently supported by the evidence: the score function and its coefficients are tuned on CIFAR-10, which is also part of the main evaluation, and no result is accompanied by error bars or multiple seeds. The reported advantage over the strongest baseline (DICE) averages only about 4% across datasets, not the 10% stated in the abstract.

major comments (4)
  1. [Appendix D and Section 5.2] The functional form f(x)=-ln(x) and the class-dependent coefficients theta_c are selected on CIFAR-10 by comparing four candidate functions on the same dataset (Figure 10), and CIFAR-10 is then part of the headline evaluation. This is a selection-on-test-data loop: the CIFAR-10 results in Figure 3 and the reported 10.18% average improvement are optimistically biased. The reported improvements over the strongest baseline DICE are 7.08% on CIFAR-10 but only 1.81% on CINIC-10 and 3.53% on DEAP, so the external evidence for the headline claim is much weaker. The authors should fix f and theta using a validation split or a separate dataset, then evaluate on held-out data, or at least present the comparison with f and theta chosen without using CIFAR-10.
  2. [Section 5] No result is accompanied by error bars, standard deviations, or the number of seeds. Figures 3-7 and the text report single runs for each configuration. Given the high variance typical of federated training with Dirichlet-distributed non-IID data, the claimed 10.18% average improvement and the smaller differences in Appendix K cannot be distinguished from stochastic variation. Please report means and standard deviations over at least three independent runs, and where possible a paired significance test.
  3. [Section 4.2 and Section 3.1] The PASS protocol's reconstruction of the global distribution via Eq. (19) relies on receiving a masked distribution Y_e from every enrolled client. If any client drops out after the key-agreement step, the residual pairwise masks do not cancel and the server cannot compute N_s; the protocol has no dropout handling. The assumptions in Section 3.1 state that participants follow the protocol, but dropout is a separate reliability issue that is common in cross-silo FL. The authors should either extend PASS with dropout resilience (e.g., via secret sharing of the masks) or explicitly discuss this limitation in the main text; the current Discussion section only covers malicious clients and free-riding.
  4. [Algorithm 1 (Section 3.3)] The payment-determination loop (lines 15-24) is not well-defined when every client in V^{-e} satisfies the budget constraint. If the for loop completes without executing break, the variable j is incremented past the last index of V^{-e} (from j=E-1 to j=E), and line 23 reads b_j and u_j out of bounds. In this case, no critical value is computed. The proofs of Lemma 3.3 and Theorem 3.7 also assume the existence of an index \hat{k} at which the budget constraint is first violated; for a sufficiently large budget R, such an index does not exist. The authors should handle this boundary case (e.g., by defining the critical payment using the budget constraint alone when all other clients pass) and update the proofs and worked example accordingly.
minor comments (4)
  1. [Section 5.3 and Abstract] The abstract claims 'outperforms the in-training baseline with more than 2% accuracy increase', but on CIFAR-10 FLMarket's average accuracy (66.12%) is lower than S-FedAvg's (66.67%); the reported 2.1% average is only an average across datasets. Please qualify the claim.
  2. [Section 4, Eq. (19)] The phrase 'reminding the global data distribution' should be 'yielding the global data distribution'.
  3. [Appendix I] 'Sharply value' should be 'Shapley value'.
  4. [Figure 2] The axis labels in Figure 2a are garbled ('Accuracy Improvement in accuracy Number of Data Accuracy'); please clean up the figure.

Circularity Check

1 steps flagged · score 4.0 of 10

The functional form f(·) and theta_c weighting are selected on CIFAR-10 in Appendix D and then CIFAR-10 is reused in the headline 10.18% accuracy comparison; auction and privacy contributions remain independent.

  1. fitted input called prediction [Appendix D (Choosing f(·) and θ_c in Evaluation Function), Eq. (1)-(2), and Section 5.2 (Comparison with Pre-training Selection)]
    "We use CIFAR-10 dataset to study the effectiveness of the score function u_e as shown in Equation 1 (§2.2). ... Figure 10b shows that the average accuracy of f0(x) = − ln(x) across six distributions is 1.71% higher than the average accuracy of the other three functions ... In summary, compared to pre-training client selection baselines, FLMarket achieves an average improvement of 10.18% in accuracy across different datasets, data distributions, and selection modes."

    The score function in Eq. (1)-(2) is not fixed a priori: Appendix D chooses f0(x) = -ln(x) among four candidates by comparing the resulting FLMarket accuracy on CIFAR-10 distributions D1-D6, and also validates varying theta_c on CIFAR-10. The main evaluation then reuses CIFAR-10 as one of the three datasets, reporting up to 40.78-47.05% gains there and folding those numbers into the aggregate 10.18% average. Thus the CIFAR-10 portion of the headline accuracy claim is a re-measurement of the same data used for model selection rather than an independent prediction. CINIC-10 and DEAP provide partial external validation, and the auction/PASS results are independent, so the circularity is partial rather than definitional.

full rationale

The only load-bearing circular step is the model-selection loop on CIFAR-10: the paper tunes f(·) and the theta_c scheme in Appendix D using CIFAR-10 accuracy, then reports CIFAR-10 accuracy as part of its main evidence and aggregates it into the 10.18% claim. This is a fitted-input-called-prediction pattern for the CIFAR-10 column, and it inflates the headline average. The auction mechanism (truthfulness, individual rationality, budget feasibility) is proven from the cited external theorem of Singer [41] and the appendix lemmas; I found no self-citation chain or uniqueness-theorem circularity. The PASS protocol's security analysis is self-contained and its main limitation (all clients must complete the protocol) is explicitly acknowledged, not hidden. CINIC-10 and DEAP results give some independent support, so the central claim is not definitionally forced. Correctness concerns such as missing error bars and single-seed runs are real but are not circularity; they do not affect this score directly. Overall circularity score: 4.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

No truly new entities; the pseudo-distribution Y_e is a masked vector, not a new object. The central claim rests on a hand-chosen score function and on strong client availability assumptions.

free parameters (1)
  • f(.) in score function = -ln(x)
    Chosen among four candidates by measured accuracy on CIFAR-10 (Appendix D). This functional choice directly determines client scores and selection.
assumptions (4)
  • domain assumption All enrolled clients complete the PASS protocol and are semi-honest.
    PASS noise cancellation in Eq. (19) requires Y_e from every client; stated in Section 3.1 and used in Theorem G.1.
  • ad hoc to paper The true value of a client's data for FL accuracy follows a diminishing-return curve approximated by -ln(x).
    Invoked to define phi(.) in Section 3.2; motivated by [2] and [17] but not proven for FL client selection.
  • standard math Diffie-Hellman key agreement and PRGs are secure against the semi-honest server and clients.
    Standard cryptographic assumption used in Section 4.1 for the PASS protocol.
  • domain assumption Clients have quasi-linear utilities and bid truthfully.
    The truthfulness proofs in Theorems 3.4 and 3.5 assume bidding behavior as in [41] and [63].

how reviews work

0 comments
Cite this review

Pith. "Pith review of FLMarket: Enabling Privacy-preserved Pre-training Data Pricing for Federated Learning." pith.science (2026). https://pith.science/paper/YZQ3J4QE

@misc{pith2026241111713,
  author       = {Pith},
  title        = {Pith review of: FLMarket: Enabling Privacy-preserved Pre-training Data Pricing for Federated Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YZQ3J4QE}},
  note         = {Machine review of arXiv:2411.11713}
}
read the original abstract

Federated Learning (FL), as a mainstream privacy-preserving machine learning paradigm, offers promising solutions for privacy-critical domains such as healthcare and finance. Although extensive efforts have been dedicated from both academia and industry to improve the vanilla FL, little work focuses on the data pricing mechanism. In contrast to the straightforward in/post-training pricing techniques, we study a more difficult problem of pre-training pricing without direct information from the learning process. We propose FLMarket that integrates a two-stage, auction-based pricing mechanism with a security protocol to address the utility-privacy conflict. Through comprehensive experiments, we show that the client selection according to FLMarket can achieve more than 10% higher accuracy in subsequent FL training compared to state-of-the-art methods. In addition, it outperforms the in-training baseline with more than 2% accuracy increase and 3x run-time speedup.

Figures

Figures reproduced from arXiv: 2411.11713 by the authors.

Figure 1
Figure 1. FLMarket high-level system overview 3 DATA MARKETPLACE PRICING FRAMEWORK FOR FEDERATED LEARNING In this section, we provide an overview of the framework with the designs of the two-phase pricing mechanism [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. CIFAR-10: 𝑛 clients selected from 20 clients D1 D2 D3 D4 D5 D6 D7 0.2 0.4 0.6 0.8 1.0 A c c u r a c y Data Distribution RS QBS DICE DDS FLMARKET (a) 5 clients selected D1 D2 D3 D4 D5 D6 D7 0.2 0.4 0.6 0.8 1.0 A c c u r a c y Data Distribution RS QBS DICE DDS FLMARKET (b) 10 clients selected D1 D2 D3 D4 D5 D6 D7 0.2 0.4 0.6 0.8 1.0 A c c u r a c y Data Distribution RS QBS DICE DDS FLMARKET (c) 15 clients selected [P… view at source ↗
Figure 4
Figure 4. CINIC-10: 𝑛 clients selected from 20 clients D1 D2 D3 D4 D5 D6 0.2 0.4 0.6 0.8 1.0 A c c u r a c y Data Distribution RS QBS DICE DDS FLMARKET (a) 5 clients selected D1 D2 D3 D4 D5 D6 0.2 0.4 0.6 0.8 1.0 A c c u r a c y Data Distribution RS QBS DICE DDS FLMARKET (b) 10 clients selected D1 D2 D3 D4 D5 D6 0.2 0.4 0.6 0.8 1.0 A c c u r a c y Data Distribution RS QBS DICE DDS FLMARKET (c) 15 clients selected [PITH_FULL_… view at source ↗
Figures from the paper (9 more)
Figure 5
Figure 5. Figure 5: DEAP: 𝑛 clients selected from 20 clients 5.2 Comparison with Pre-training Selection We evaluate the performance of FLMarket on selecting clients by comparing the proposed algorithm with four pre-training client selection baselines. Figures 3, 4 and 5 show the final tes…
Figure 6
Figure 6. Figure 6: The accuracy and loss curves for FLMarket and other four baselines under the D1 distribution with a selec￾tion of 10 out of 20 clients on CIFAR-10 dataset. For the most imbalanced distributions (D4, D5, and D6), FLMar￾ket achieves the highest accuracy in 22 out of 27 t…
Figure 7
Figure 7. Figure 7: Comparison of performance between FLMarket and S-FedAvg across three datasets In summary, compared to pre-training client selection baselines, FLMarket achieves an average improvement of 10.18% in accuracy [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: FL training architecture resources to train a global model. Obviously, the training process will consume your personal resources, such as on-device compu￾tation and data communication. As compensation, the server will also offer rewards to the participating clients. FL…
Figure 9
Figure 9. Figure 9: Results of the questionnaire pricing has a positive impact (either significant or some impact) on their decision to participate in FL. These findings highlight that ensuring privacy and offering rea￾sonable rewards are crucial to enhancing participation in FL train￾ing…
Figure 10
Figure 10. Figure 10: The impact of different category coefficients [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 11
Figure 11. Figure 11: The example of PASS 1 < 3, so Client1 adds 1 · [1, 9, 7] and 1 · [2, 5, 3] generates pseudo data distribution [6, 20, 18] = [3, 6, 8] + [1, 9, 7] + [2, 5, 3]. Similarly, Client2 adds −1 · [1, 9, 7] and 1 · [2, 5, 3] to generates pseudo data distribution [8, 2, 1] = [4…
Figure 12
Figure 12. Figure 12: The generated data distributions of CIFAR-10 and an example of its allocation [PITH_FULL_IMAGE:figures/full_fig_p015_12.png]
Figure 13
Figure 13. Figure 13: CIFAR-10: 𝑛 clients selected from 100 clients To evaluate the performance of FLMarket with a large number of clients, we present the results obtained from experiments in which 25, 50, and 75 clients were selected from a pool of 100 clients. The global data distributio…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

68 extracted references · 49 canonical work pages

  1. [1]

    Priyam Basu, Tiasa Singha Roy, Rakshit Naidu, and Zumrut Muftuoglu. 2021. Privacy enabled Financial Text Classification using Differential Privacy and Federated Learning. In Proceedings of the Third Workshop on Economics and Natural Language Processing. Association for Computational Linguistics, Punta Cana, Dominican Republic, 50–55. https://doi.org/10.18...

    work page doi:10.18653/v1/2021.econlp-1.7 2021

  2. [2]

    Shai Ben-David, John Blitzer, Koby Crammer, Alex Kulesza, Fernando Pereira, and Jennifer Wortman Vaughan. 2010. A theory of learning from different domains. Machine learning 79 (2010), 151–175

    work page 2010

  3. [3]

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

    2017

  4. [4]

    Michelle Chen and Olga Ohrimenko. 2023. Protecting global properties of datasets with distribution privacy mechanisms. In International Conference on Artificial Intelligence and Statistics. PMLR, 7472–7491

    work page 2023

  5. [5]

    Luke N Darlow, Elliot J Crowley, Antreas Antoniou, and Amos J Storkey. 2018. Cinic-10 is not imagenet or cifar-10. arXiv:1810.03505 (2018)

    arXiv 2018

  6. [6]

    Ittai Dayan, Holger R Roth, Aoxiao Zhong, Ahmed Harouni, et al. 2021. Federated learning for predicting clinical outcomes in patients with COVID-19. Nature medicine 27, 10 (2021), 1735–1743

    work page 2021

  7. [7]

    Yongheng Deng, Feng Lyu, Ju Ren, Yi-Chao Chen, Peng Yang, Yuezhi Zhou, and Yaoxue Zhang. 2021. Fair: Quality-aware federated learning with precise user incentive and model aggregation. In IEEE INFOCOM 2021-IEEE Conference on Computer Communications. IEEE, 1–10

    work page 2021

  8. [8]

    Yongheng Deng, Feng Lyu, Ju Ren, Yi-Chao Chen, Peng Yang, Yuezhi Zhou, and Yaoxue Zhang. 2022. Improving federated learning with quality-aware user incentive and auto-weighted model aggregation. IEEE Transactions on Parallel and Distributed Systems 33, 12 (2022), 4515–4529

    work page 2022

Show all 68 references
  1. [9]

    Yongheng Deng, Feng Lyu, Ju Ren, Huaqing Wu, Yuezhi Zhou, Yaoxue Zhang, and Xuemin Shen. 2021. Auction: Automated and quality-aware client selection framework for efficient federated learning. IEEE Transactions on Parallel and Distributed Systems 33, 8 (2021), 1996–2009

    2021

  2. [10]

    Mukund Deshpande and George Karypis. 2004. Item-based top-n recommenda- tion algorithms. ACM Transactions on Information Systems (TOIS) 22, 1 (2004), 143–177

    2004

  3. [11]

    Liang Gao, Li Li, Yingwen Chen, Wenli Zheng, ChengZhong Xu, and Ming Xu

  4. [12]

    Valentin Hartmann, Léo Meynent, Maxime Peyrard, Dimitrios Dimitriadis, Shruti Tople, and Robert West. 2023. Distribution inference risks: Identifying and mitigating sources of leakage. In 2023 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML). IEEE, 136–149

    2023

  5. [13]

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2016. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition . 770–778

    2016

  6. [14]

    Martin Hellman. 1976. New directions in cryptography. IEEE transactions on Information Theory 22, 6 (1976), 644–654

    1976

  7. [15]

    Tzu-Ming Harry Hsu, Hang Qi, et al. 2019. Measuring the effects of non-identical data distribution for federated visual classification. arXiv:1909.06335 (2019)

    2019 arXiv

  8. [16]

    Miao Hu, Di Wu, Yipeng Zhou, Xu Chen, and Min Chen. 2022. Incentive-aware autonomous client participation in federated learning. IEEE Transactions on Parallel and Distributed Systems 33, 10 (2022), 2612–2627

    2022

  9. [17]

    Jared Kaplan, Sam McCandlish, Tom Henighan, Tom B Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei. 2020. Scaling laws for neural language models. arXiv preprint arXiv:2001.08361 (2020)

    2020 arXiv

  10. [18]

    Sander Koelstra, Christian Muhl, Mohammad Soleymani, Jong-Seok Lee, Ashkan Yazdani, Touradj Ebrahimi, Thierry Pun, Anton Nijholt, and Ioannis Patras. 2011. Deap: A database for emotion analysis; using physiological signals. IEEE transac- tions on affective computing 3, 1 (2011), 18–31

    2011

  11. [19]

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

    2009

  12. [20]

    Fan Lai, Xiangfeng Zhu, Harsha V Madhyastha, and Mosharaf Chowdhury

  13. [21]

    Anran Li, Lan Zhang, Juntao Tan, Yaxuan Qin, Junhao Wang, and Xiang-Yang Li. 2021. Sample-level data selection for federated learning. In IEEE INFOCOM 2021-IEEE Conference on Computer Communications . IEEE, 1–10

    2021

  14. [22]

    In 15th{USENIX} Symposium on Operating Systems Design and Implementation ({OSDI} 21)

    Oort: Efficient federated learning via guided participant selection. In 15th{USENIX} Symposium on Operating Systems Design and Implementation ({OSDI} 21). 19–35

  15. [23]

    Qi Li, Zhuotao Liu, Qi Li, and Ke Xu. 2023. martFL: Enabling Utility-Driven Data Marketplace with a Robust and Verifiable Federated Learning Architecture. In Proceedings of the 2023 ACM SIGSAC Conference on Computer and Communications Security. 1496–1510

    2023

  16. [24]

    Qinbin Li, Yiqun Diao, Quan Chen, and Bingsheng He. 2022. Federated learning on non-iid data silos: An experimental study. In 2022 IEEE 38th International Conference on Data Engineering (ICDE) . IEEE, 965–978

    2022

  17. [25]

    Zitao Li, Bolin Ding, Liuyi Yao, Yaliang Li, Xiaokui Xiao, and Jingren Zhou. 2024. Performance-Based Pricing for Federated Learning via Auction. Proceedings of the VLDB Endowment 17, 6 (2024), 1269–1282

    2024

  18. [26]

    Tian Li, Anit Kumar Sahu, Manzil Zaheer, Maziar Sanjabi, Ameet Talwalkar, and Virginia Smith. 2020. Federated optimization in heterogeneous networks. Proceedings of Machine learning and systems 2 (2020), 429–450

    2020

  19. [27]

    Tao Lin, Lingjing Kong, Sebastian U Stich, and Martin Jaggi. 2020. Ensemble distillation for robust model fusion in federated learning. Advances in Neural Information Processing Systems 33 (2020), 2351–2363

    2020

  20. [28]

    Jierui Lin, Min Du, and Jian Liu. 2019. Free-riders in federated learning: Attacks and defenses. arXiv:1911.12560 (2019)

    2019 arXiv

  21. [29]

    Yuan Liu, Zhengpeng Ai, Shuai Sun, Shuangfeng Zhang, Zelei Liu, and Han Yu. 2020. Fedcoin: A peer-to-peer payment system for federated learning. In Federated Learning. Springer, 125–138

    2020

  22. [30]

    Yaguang Lin, Zhipeng Cai, Xiaoming Wang, Fei Hao, Liang Wang, and Akshita Maradapu Vera Venkata Sai. 2021. Multi-round incentive mechanism for cold KDD ’25, August 03–07,2025, Toronto,Canada Zhenyu Wen, Wanglei Feng, Di Wu, Haozhen Hu, Chang Xu, Bin Qian, Zhen Hong, Cong Wang,...

    2021

  23. [31]

    Wuxing Mao, Qian Ma, Guocheng Liao, and Xu Chen. 2024. Game Analysis and Incentive Mechanism Design for Differentially Private Cross-silo Federated Learning. IEEE Transactions on Mobile Computing (2024)

    2024

  24. [32]

    Ziyang Liu and Hakan Hacigümüs. 2014. Online optimization and fair costing for dynamic data sharing in a cloud data market. In Proceedings of the 2014 ACM SIGMOD International Conference on Management of Data . 1359–1370

    2014

  25. [33]

    Lokesh Nagalapatti and Ramasuri Narayanam. 2021. Game of gradients: Mitigat- ing irrelevant clients in federated learning. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 35. 9046–9054

    2021

  26. [34]

    R Preston McAfee and John McMillan. 1987. Auctions and bidding. Journal of economic literature 25, 2 (1987), 699–738

    1987

  27. [35]

    Jake Perazzone, Shiqiang Wang, Mingyue Ji, and Kevin S Chan. 2022. Communication-efficient device scheduling for federated learning using sto- chastic optimization. In IEEE INFOCOM 2022-IEEE Conference on Computer Com- munications. IEEE, 1449–1458

    2022

  28. [36]

    Takayuki Nishio and Ryo Yonetani. 2019. Client selection for federated learning with heterogeneous resources in mobile edge. InICC 2019-2019 IEEE international conference on communications (ICC) . IEEE, 1–7

    2019

  29. [37]

    Kean Ren, Guocheng Liao, Qian Ma, and Xu Chen. 2023. Differentially Private Auction Design for Federated Learning with non-IID Data. IEEE Transactions on Services Computing (2023), 1–12

    2023

  30. [38]

    Bin Qian, Jie Su, Zhenyu Wen, Devki Nandan Jha, Yinhao Li, Yu Guan, Deepak Puthal, Philip James, Renyu Yang, Albert Y Zomaya, et al. 2020. Orchestrating the development lifecycle of machine learning-based IoT applications: A taxonomy and survey. ACM Computing Surveys (CSUR) 53...

    2020

  31. [39]

    Rituparna Saha, Sudip Misra, Aishwariya Chakraborty, Chandranath Chatterjee, and Pallav Kumar Deb. 2022. Data-Centric Client Selection for Federated Learning Over Distributed Edge Networks. IEEE Transactions on Parallel and Distributed Systems 34, 2 (2022), 675–686

    2022

  32. [40]

    Monica Ribero and Haris Vikalo. 2020. Communication-efficient federated learn- ing via optimal client sampling. arXiv:2007.15197 (2020)

    2020 arXiv

  33. [41]

    Yaron Singer. 2010. Budget feasible mechanisms. In 2010 IEEE 51st Annual Sym- posium on foundations of computer science . IEEE, 765–774

    2010

  34. [42]

    Fabian Schomm, Florian Stahl, and Gottfried Vossen. 2013. Marketplaces for data: an initial survey. ACM SIGMOD Record 42, 1 (2013), 15–26

    2013

  35. [43]

    Sarah Spiekermann, Alessandro Acquisti, Rainer Böhme, and Kai-Lung Hui. 2015. The challenges of personal data markets and privacy.Electronic markets 25 (2015), 161–167

    2015

  36. [44]

    Tianshu Song et al. 2019. Profit allocation for federated learning. In 2019 IEEE International Conference on Big Data (Big Data) . IEEE, 2577–2586

    2019

  37. [45]

    Peng Sun, Guocheng Liao, Xu Chen, and Jianwei Huang. 2024. A Socially Optimal Data Marketplace With Differentially Private Federated Learning. IEEE/ACM Transactions on Networking (2024)

    2024

  38. [46]

    Peng Sun, Xu Chen, Guocheng Liao, and Jianwei Huang. 2022. A profit- maximizing model marketplace with differentially private federated learning. In IEEE INFOCOM 2022-IEEE Conference on Computer Communications . IEEE, 1439–1448

    2022

  39. [47]

    Xiaole Wan, Dongqian Yang, Tongtong Wang, and Muhammet Deveci. 2023. Closed-loop supply chain decision considering information reliability and secu- rity: should the supply chain adopt federated learning decision support systems? Annals of Operations Research (2023), 1–37

    2023

  40. [48]

    Ming Tang and Vincent WS Wong. 2021. An incentive mechanism for cross-silo federated learning: A public goods perspective. In IEEE INFOCOM 2021-IEEE Conference on Computer Communications . IEEE, 1–10

    2021

  41. [49]

    Guan Wang, Charlie Xiaoqian Dang, and Ziye Zhou. 2019. Measure contribution of participants in federated learning. In 2019 IEEE International Conference on Big Data (Big Data). IEEE, 2597–2604

    2019

  42. [50]

    Cong Wang, Yuanyuan Yang, and Pengzhan Zhou. 2021. Towards Efficient Scheduling of Federated Mobile Devices Under Computational and Statistical Heterogeneity. IEEE Transactions on Parallel and Distributed Systems 32, 2 (2021), 394–410. https://doi.org/10.1109/TPDS.2020.3023905

    2021

  43. [51]

    Mišić, and Yixiang Wang

    Jianhua Wang, Xiaolin Chang, Jelena Mišić, Vojislav B. Mišić, and Yixiang Wang

  44. [52]

    Hongyi Wang, Mikhail Yurochkin, Yuekai Sun, Dimitris Papailiopoulos, and Yasaman Khazaeni. 2020. Federated learning with matched averaging. arXiv preprint arXiv:2002.06440 (2020)

    2020 arXiv

  45. [53]

    Qinyong Wang, Hongzhi Yin, Tong Chen, Junliang Yu, Alexander Zhou, and Xiangliang Zhang. 2021. Fast-adapting and privacy-preserving federated recom- mender system. The VLDB Journal 31, 5 (oct 2021), 877–896

    2021

  46. [54]

    Xinyi Xu, Lingjuan Lyu, Xingjun Ma, Chenglin Miao, Chuan Sheng Foo, and Bryan Kian Hsiang Low. 2021. Gradient driven rewards to guarantee fairness in collaborative machine learning. Advances in Neural Information Processing Systems 34 (2021), 16104–16117

    2021

  47. [55]

    Junxiao Wang, Song Guo, Xin Xie, and Heng Qi. 2022. Protect privacy from gradi- ent leakage attack in federated learning. In IEEE INFOCOM 2022-IEEE Conference on Computer Communications. IEEE, 580–589

    2022

  48. [56]

    Rongfei Zeng, Chao Zeng, Xingwei Wang, Bo Li, and Xiaowen Chu. 2022. In- centive Mechanisms in Federated Learning and A Game-Theoretical Approach. IEEE Network 36, 6 (2022), 229–235

    2022

  49. [57]

    Rongfei Zeng, Shixun Zhang, Jiaqi Wang, and Xiaowen Chu. 2020. Fmore: An incentive scheme of multi-dimensional auction for federated learning in mec. In 2020 IEEE 40th International Conference on Distributed Computing Systems (ICDCS). IEEE, 278–288

    2020

  50. [58]

    Andrew C Yao. 1982. Theory and application of trapdoor functions. In 23rd Annual Symposium on Foundations of Computer Science (SFCS 1982) . IEEE, 80–91

    1982

  51. [59]

    Yufeng Zhan, Jie Zhang, Zicong Hong, Leijie Wu, Peng Li, and Song Guo. 2021. A survey of incentive mechanism design for federated learning. IEEE Transactions on Emerging Topics in Computing 10, 2 (2021), 1035–1044

    2021

  52. [60]

    Jingwen Zhang, Yuezhou Wu, and Rong Pan. 2021. Incentive mechanism for hori- zontal federated learning based on reputation and reverse auction. In Proceedings of the Web Conference 2021. 947–956

    2021

  53. [61]

    Yufeng Zhan, Peng Li, Zhihao Qu, Deze Zeng, and Song Guo. 2020. A learning- based incentive mechanism for federated learning.IEEE Internet of Things Journal 7, 7 (2020), 6360–6368

    2020

  54. [62]

    Shuyuan Zheng, Yang Cao, Masatoshi Yoshikawa, Huizhong Li, and Qiang Yan

  55. [63]

    Zhenzhe Zheng, Fan Wu, Xiaofeng Gao, Hongzi Zhu, Shaojie Tang, and Guihai Chen. 2016. A budget feasible incentive mechanism for weighted coverage maximization in mobile crowdsensing. IEEE Transactions on Mobile Computing 16, 9 (2016), 2392–2407

    2016

  56. [64]

    Wanrong Zhang, Shruti Tople, and Olga Ohrimenko. 2021. Leakage of dataset properties in{Multi-Party} machine learning. In30th USENIX security symposium (USENIX Security 21). 2687–2704

    2021

  57. [68]

    Ruiting Zhou, Jinlong Pang, Zhibo Wang, John CS Lui, and Zongpeng Li. 2021. A truthful procurement auction for incentivizing heterogeneous clients in federated learning. In 2021 IEEE 41st International Conference on Distributed Computing Systems (ICDCS). IEEE, 183–193. APPENDI...

    2021

  58. [2021]

    In Proceedings of the 50th International Conference on Parallel Processing

    Fifl: A fair incentive mechanism for federated learning. In Proceedings of the 50th International Conference on Parallel Processing . 1–10

  59. [2022]

    In 2022 IEEE International Conference on Big Data (Big Data)

    FL-Market: Trading private models in federated learning. In 2022 IEEE International Conference on Big Data (Big Data) . IEEE, 1525–1534

    2022

  60. [2024]

    IEEE Internet of Things Journal 11, 1 (2024), 1374–1384

    PASS: A Parameter Audit-Based Secure and Fair Federated Learning Scheme Against Free-Rider Attack. IEEE Internet of Things Journal 11, 1 (2024), 1374–1384

    2024

Pith tools

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