Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

FedCostAware: Enabling Cost-Aware Federated Learning on the Cloud

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read FedCostAware turns the idle wait in synchronous federated learning into a cost-saving strategy on cloud spot instances, cutting GPU bills by up to 72% compared with on-demand instances.

desk verdict A cost-aware spot-instance scheduler for synchronous FL that delivers real savings in a benign cloud window, though the fault-tolerance machinery that justifies spot use is never exercised. read the letter →

arxiv 2505.21727 v2 pith:XEUWOIVT submitted 2025-05-27 cs.DC

Aditya Sinha , Zilinghan Li , Tingkai Liu , Volodymyr Kindratenko , Kibaek Kim , Ravi Madduri This is my paper
classification cs.DC
keywords federatedlearningcloudcomputingspotinstancescostoptimizationschedulingstragglerproblemsynchronousFLinstancelifecyclemanagement
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 claims that the idle time inherent to synchronous federated learning—fast clients waiting on stragglers—can be converted into direct cost savings on cloud spot instances. It introduces FedCostAware, a scheduler that terminates a client's spot instance as soon as its local training finishes and pre-warms the instance just before the next round is expected to start, using per-client estimates of epoch time and spin-up time. The authors report that this reduces federated training cost by roughly 60% compared to always-on spot instances and up to 72.22% compared to on-demand instances across four datasets. The significance is that synchronous FL, which is preferred for model accuracy and stability, can become affordable on volatile, discount-priced cloud capacity instead of requiring expensive on-demand GPUs.

What carries the argument

The termination-and-prewarm scheduler, driven by a calibration phase and exponential moving average updates, is the central mechanism. During the first two rounds the system measures each client's cold-start epoch time, warm-start epoch time, and instance spin-up time. Thereafter, whenever a client finishes early, the scheduler estimates the round's slowest finish time, computes the idle window, and if the idle window exceeds the spin-up time plus a threshold, it terminates the instance and schedules a pre-warm at the estimated slowest finish minus spin-up minus a buffer. The same estimates are updated online with an exponential moving average, and a separate re-estimation step adjusts the pre-warm schedule when a spot preemption forces a crashed client to recover from a checkpoint.

What would settle it

Run the same federated workload on spot instances while deliberately doubling the observed spin-up times or forcing preemptions in the middle of a round; if the wall-clock time per round grows or the total cost exceeds the always-on spot baseline, the central claim that lifecycle management reliably reduces cost would be refuted.

Watch

Extended reading notes

Core claim

The central claim is that a cost-aware scheduler can make synchronous federated learning on spot instances cheaper than both conventional spot usage and on-demand usage while leaving the aggregation protocol unchanged. The mechanism is to treat the wait for the slowest client as billable idle time and eliminate it: each client that finishes early has its instance terminated, and a pre-warming queue restarts it at a time computed from the estimated slowest finish time minus the estimated spin-up time, with a buffer to avoid delaying the next round. The paper demonstrates on MNIST, CIFAR-10, AI-READI, and Fed-ISIC2019 that this lifecycle management yields total-cost reductions of 67–72% relative to on-demand instances and about 6–11 percentage points of additional savings relative to naive spot usage, with negligible added wall-clock delay (at most about 3 minutes per experiment from spin-up estimation error).

Load-bearing premise

The cost savings depend on the assumption that instance spin-up time can be predicted accurately enough that terminating and restarting an instance is cheaper than keeping it idle; if boot times are volatile or spot instances are unavailable when pre-warming starts, synchronous rounds will stall and the savings vanish.

Editorial extensions

If this is right

  • If the reported savings hold, synchronous federated learning can be deployed on spot instances at roughly 30–40% the cost of on-demand deployment, making GPU-based FL feasible for budget-constrained institutions such as hospitals.
  • The approach preserves the synchronous aggregation protocol, so existing convergence guarantees and model quality of synchronous FL are retained rather than traded away for cost.
  • Because the savings scale linearly with the hourly GPU price, the same scheduler would yield proportionally larger dollar savings on high-end accelerators ($30–$100 per hour).
  • Client-defined budgets become enforceable: a client whose remaining budget cannot cover the next round is automatically excluded, so no participant exceeds its financial limit.
  • Checkpoint-and-resume makes long FL runs resilient to spot preemptions, provided the pre-warming schedule is updated to the crashed client's new estimated finish time.

Reading between the lines

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

  • The paper's experiments saw no spot preemptions, so the dynamic schedule adjustment and checkpoint recovery are untested under real interruptions; a natural next step is to force preemptions during training and measure whether cost savings survive the recovery overhead.
  • The same terminate-and-prewarm pattern applies beyond federated learning to any synchronous distributed job with stragglers, such as data-parallel deep learning with heterogeneous workers, so long as per-worker completion times can be predicted.
  • The fixed threshold and EMA estimates could be replaced by a learned predictor of boot time and spot price, potentially recovering savings in settings where spin-up time is volatile or spot prices fluctuate within a round.
  • If cloud providers move to finer-grained billing, the idle periods being eliminated may already be cheap, which would reduce the advantage of termination over simply idling; the scheduler's benefit is largest under per-hour billing, which is the current norm.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes FedCostAware, a scheduler for synchronous federated learning that runs client training on cloud spot instances and reduces cost by terminating instances during idle periods caused by stragglers, then pre-warming them before the next round based on estimated spin-up times. The system also performs checkpointing for fault tolerance, dynamically adjusts schedules after preemptions, and enforces per-client budgets. Experiments on Fed-ISIC2019, AI-READI, CIFAR-10, and MNIST report cost savings of 60-72% over on-demand instances and roughly 25% over a static spot-instance baseline, while preserving the synchronous aggregation protocol.

Significance. If the central claim holds, FedCostAware addresses a real operational cost in cross-silo federated learning: idle GPU time during synchronous rounds. The natural-partition experiments on Fed-ISIC2019 and AI-READI are a genuine strength, since they do not rely on artificially injected delays. The cost measurements are taken directly from billed instance time, so the reported savings are not an artifact of a fitted parameter or a normalization convention. The paper is also honest about the absence of preemptions in its experiments. However, the evaluation is incomplete for a system whose main selling point is cost-effective use of volatile spot instances, because the fault-tolerance and dynamic-schedule-adjustment paths are never exercised, and the paper provides no accuracy results or repeated trials.

major comments (4)
  1. [Section IV-B and Section III-D] The central empirical claim is tested only in a preemption-free window. Section IV-B states that no spot preemptions occurred, even in sessions exceeding six hours, so the checkpoint-recovery and dynamic schedule adjustment mechanisms described in Section III-D were never exercised. Since spot preemption is the defining failure mode of the target resource class, the paper needs fault-injection experiments in which instances are terminated mid-round at controlled rates; the evaluation should report recovery overhead, round-delay impact, and total cost under varying preemption frequencies. Without such experiments, the claim that FedCostAware 'significantly reduces cloud computing costs' is established only for a benign spot market, not for the volatile conditions the system is designed to handle.
  2. [Section IV-A and Table I] All cost numbers in Table I come from single runs without error bars or repeated trials. Spot prices, boot times, and network conditions vary over time and across regions, so a single measurement cannot support the claimed savings percentages as stable quantities. The paper should report the mean and spread over multiple runs, and ideally over more than one AWS region or availability zone, to support the generality of the cost-savings claim.
  3. [Section IV-B and Table I] The paper reports only training time and cost, never model accuracy or convergence. Because FedCostAware keeps the synchronous aggregation protocol, one expects accuracy to match a standard synchronous baseline, but the paper does not demonstrate this. A practical cost-saving system must show that the savings do not degrade the trained model; the authors should add final accuracy or loss per dataset for FedCostAware, the static spot baseline, and the on-demand baseline.
  4. [Section IV-A and Table I] The headline savings of 72.22% come from CIFAR-10, where per-epoch times are artificially scaled by client-specific factors, while the most realistic datasets show lower savings (70.47% and 67.18%). The paper does not state how these scaling factors were chosen or whether they were selected to create a particular straggler spread. Since the magnitude of idle-time savings depends directly on the spread of client finishing times, the authors should report the scaling factors, justify them, and present a sensitivity analysis showing how savings vary with the degree of heterogeneity.
minor comments (5)
  1. [Listing 1 and Section III-B] The notation is inconsistent: the text uses T_epoch cold and T_epoch warm, while Listing 1 uses T_epoch_cold and T_epoch_warm; also F_s is used both for the estimated slowest finish time inside the function and for the slowest client's finish time in the surrounding text. Please unify the notation.
  2. [Table I] The 'Savings (%)' column does not specify the baseline for each row. For the two Sync FL rows the savings are relative to on-demand, and for the FedCostAware row they are also relative to on-demand, but the column header alone is ambiguous. Please add a footnote or clarify the comparison baseline.
  3. [Abstract and Section V] The abstract states 'up to a 70% reduction' while the conclusion and Table I report 72.22%; this inconsistency should be corrected.
  4. [Section IV-B] The sentence about P4 ($30/hr) and P5 ($100/hr) GPUs is speculative; the paper does not test these instance types. It would be better phrased as an expectation rather than a demonstrated scaling result.
  5. [Section III-A] The phrase 'all datasets remain within the secure cloud accounts of the respective clients' is slightly ambiguous given that the system provisions instances across providers and accounts; the subsequent explanation via S3 presigned URLs for model updates should be integrated more clearly.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the cost-savings claim is measured directly from billed instance time, and the calibration and self-citations do not force the result.

full rationale

The central claim, that FedCostAware reduces cloud computing costs for synchronous FL, is an empirical result. Table I reports actual training time, instance price, and total billed cost for FedCostAware versus synchronous FL on spot and on-demand instances. The savings are computed from observed billed instance-hours, not from a fitted model: the calibration of T_epoch_cold, T_epoch_warm, and T_spin_up (Section III-B) feeds the termination/prewarming heuristic in Listing 1, but the reported costs are the observed costs of the runs, so the conclusion does not reduce to the calibration by construction. The termination rule (terminate if idle_time - T_spin_up > T_threshold) is a design choice; its claimed benefit is verified by the measured cost difference, not assumed by definition. The self-citations (APPFL [30,31], FedCompass [11], APPFLx [6], etc.) are implementation dependencies or related-work references and are not used to justify the cost-savings claim. The stated limitation, 'No spot preemptions occurred during our experiments, even in long-running sessions exceeding six hours' (Section IV-B), weakens the generality of the fault-tolerance and dynamic-schedule-adjustment claims, but that is an empirical coverage limitation, not a circularity. No equation or definition makes the claimed savings equivalent to its inputs, and no fitted parameter is renamed as a prediction.

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

The central claim rests on the assumption that terminating and restarting instances costs less than keeping them idle, plus the unverified assumption that synchronous convergence is unchanged. There are no new physical entities or mediators.

free parameters (4)
  • T_threshold = not reported
    Idle time threshold in Listing 1 that decides when to terminate an instance; hand-chosen, no value or sensitivity analysis given.
  • T_buffer = not reported
    Safety margin added to prewarm start time to avoid delays; hand-chosen, value not reported.
  • EMA smoothing factor = not reported
    Exponential moving average factor used to update T_epoch_cold, T_epoch_warm, and T_spinup; chosen without justification.
  • Client-specific scaling factors = not reported
    Introduced for MNIST and CIFAR-10 to simulate longer training times; these factors determine the straggler patterns and are not derived from measured workloads.
assumptions (4)
  • domain assumption Spot instance billing stops immediately upon termination.
    The entire savings model depends on terminating instances during idle periods to stop billing. The paper assumes AWS per-second billing without stating it explicitly.
  • domain assumption Instance spin-up time is predictable from calibration and EMA updates.
    Listing 1 uses T_spinup to schedule prewarming. If spot availability or boot time is volatile, the prewarming buffer may fail and synchronous rounds stall.
  • domain assumption Synchronous FL convergence is unaffected by lifecycle management.
    The paper claims preservation of synchronous convergence properties but reports no accuracy comparison against baselines. The claim is asserted, not demonstrated.
  • domain assumption Cloud storage and data transfer costs are negligible.
    The paper states S3 transfer costs are negligible, but the cost model excludes them. On large models or frequent checkpointing this may not hold.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FedCostAware: Enabling Cost-Aware Federated Learning on the Cloud." pith.science (2026). https://pith.science/paper/XEUWOIVT

@misc{pith2026250521727,
  author       = {Pith},
  title        = {Pith review of: FedCostAware: Enabling Cost-Aware Federated Learning on the Cloud},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XEUWOIVT}},
  note         = {Machine review of arXiv:2505.21727}
}
read the original abstract

Federated learning (FL) is a distributed machine learning (ML) approach that allows multiple clients to collaboratively train ML models without exchanging original training data, offering a solution that is particularly valuable in sensitive domains such as biomedicine. However, training robust FL models often requires substantial computing resources from participating clients, which may not be readily available at institutions such as hospitals. While cloud platforms offer on-demand access to such resources, their usage can incur significant costs, particularly in distributed training scenarios where poor coordination strategies can lead to substantial resource wastage. To address this, we introduce FedCostAware, a cost-aware scheduling algorithm designed to optimize synchronous FL on cloud spot instances. FedCostAware addresses the challenges of training on spot instances and different client budgets by employing intelligent management of the lifecycle of spot instances. This approach minimizes resource idle time and overall expenses. Comprehensive experiments across multiple datasets demonstrate that FedCostAware significantly reduces cloud computing costs compared to conventional spot and on-demand schemes, enhancing the accessibility and affordability of FL.

Figures

Figures reproduced from arXiv: 2505.21727 by the authors.

Figure 1
Figure 1. System Architecture of the Cost-Aware Federated Learning System. The central server coordinates the process with the FL server for model aggregation, the Ray head node for cluster management, and the FedCostAware scheduler for cost optimization. The central server launches/terminates client instances, initiates training, receives updates, and handles model transfers to/from the cloud storage. Client nodes are spot i… view at source ↗
Figure 2
Figure 2. Comparison of Client Activities per Epoch Between Standard and Cost-Aware Synchronous FL Processes. In standard synchronous FL, significant GPU times are wasted for faster clients while waiting for the slowest client. The proposed cost-aware scheduling algorithm converts this idle time into cost savings by stopping instances and proactively restarting them just before the next training round begins, synchronized wit… view at source ↗
Figure 3
Figure 3. The System’s Fault Tolerance Mechanism. Recov￾ering from spot instance interruptions by checkpointing and schedule adjustment [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Client operational states over time for the Fed [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Sign in 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. FedStrategist: A Meta-Learning Framework for Adaptive and Robust Aggregation in Federated Learning

    cs.LG 2025-07 reject novelty 4.0 of 10

    A LinUCB contextual bandit selects federated aggregation rules online based on update variance, cosine similarity, and mean norm, claiming superior accuracy and tunable risk posture.

Reference graph

Works this paper leans on

37 extracted references · 23 canonical work pages · cited by 1 Pith paper

  1. [1]

    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,”Artificial intelligence and statistics, pp. 1273–1282, 2017

    2017

  2. [2]

    Federated learning: Challenges, methods, and future directions,

    T. Li, A. K. Sahu, A. Talwalkar, and V . Smith, “Federated learning: Challenges, methods, and future directions,”IEEE Signal Processing Magazine, vol. 37, no. 3, pp. 50–60, 2020

    2020

  3. [3]

    Privacy-preserving federated learning for science: Challenges and research directions,

    K. Kim, K. Raghavan, O. Kotevska, M. Dorier, R. Madduri, M. Ryu, T. Munson, R. Ross, T. Flynn, A. Kagawaet al., “Privacy-preserving federated learning for science: Challenges and research directions,” in 2024 IEEE International Conference on Big Data (BigData). IEEE, 2024, pp. 7849–7853

    work page 2024

  4. [4]

    End-to-end privacy preserving deep learning on multi-institutional medical imaging,

    G. Kaissis, A. Ziller, J. Passerat-Palmbach, T. Ryffel, D. Usynin, A. Trask, I. Lima Jr, J. Mancuso, F. Jungmann, M.-M. Steinbornet al., “End-to-end privacy preserving deep learning on multi-institutional medical imaging,”Nature Machine Intelligence, vol. 3, no. 6, pp. 473– 484, 2021

    work page 2021

  5. [5]

    Federated learning enables big data for rare cancer boundary detection,

    S. Pati, U. Baid, B. Edwards, M. Sheller, S.-H. Wang, G. A. Reina, P. Foley, A. Gruzdev, D. Karkada, C. Davatzikoset al., “Federated learning enables big data for rare cancer boundary detection,”Nature Communications, vol. 13, no. 1, p. 7346, 2022

    work page 2022

  6. [6]

    Enabling End-to-End Secure Federated Learning in Biomedical Research on Heterogeneous Computing Environments with APPFLx

    T.-H. Hoang, J. Fuhrman, R. Madduri, M. Li, P. Chaturvedi, Z. Li, K. Kim, M. Ryu, R. Chard, E. Huertaet al., “Enabling end-to-end secure federated learning in biomedical research on heterogeneous computing environments with APPFLx,”arXiv preprint arXiv:2312.08701, 2023

    work page Pith review arXiv 2023

  7. [7]

    Federated learning in a medical context: a systematic literature review,

    B. Pfitzner, N. Steckhan, and B. Arnrich, “Federated learning in a medical context: a systematic literature review,”ACM Transactions on Internet Technology (TOIT), vol. 21, no. 2, pp. 1–31, 2021

    work page 2021

  8. [8]

    Heterogeneous feder- ated learning: State-of-the-art and research challenges,

    M. Ye, X. Fang, B. Du, P. C. Yuen, and D. Tao, “Heterogeneous feder- ated learning: State-of-the-art and research challenges,”ACM Computing Surveys, vol. 56, no. 3, pp. 1–44, 2023

    2023

Show all 37 references
  1. [9]

    Asynchronous federated optimization,

    C. Xie, S. Koyejo, and I. Gupta, “Asynchronous federated optimization,” arXiv preprint arXiv:1903.03934, 2019

    1903 arXiv

  2. [10]

    Federated learning with buffered asynchronous aggregation,

    J. Nguyen, K. Malik, H. Zhan, A. Yousefpour, M. Rabbat, M. Malek, and D. Huba, “Federated learning with buffered asynchronous aggregation,” inInternational conference on artificial intelligence and statistics. PMLR, 2022, pp. 3581–3607

    2022

  3. [11]

    FedCompass: Efficient cross-silo federated learning on heterogeneous client devices using a computing power-aware scheduler,

    Z. Li, P. Chaturvedi, S. He, H. Chen, G. Singh, V . Kindratenko, E. A. Huerta, K. Kim, and R. Madduri, “FedCompass: Efficient cross-silo federated learning on heterogeneous client devices using a computing power-aware scheduler,” inThe Twelfth International Conference on Learn...

    2024

  4. [12]

    Asynchronous federated stochastic optimiza- tion for heterogeneous objectives under arbitrary delays,

    C. Iakovidou and K. Kim, “Asynchronous federated stochastic optimiza- tion for heterogeneous objectives under arbitrary delays,”arXiv preprint arXiv:2405.10123, 2024

    2024

  5. [13]

    Analysis and evaluation of synchronous and asynchronous flchain,

    F. Wilhelmi, L. Giupponi, and P. Dini, “Analysis and evaluation of synchronous and asynchronous flchain,” 2022. [Online]. Available: https://arxiv.org/abs/2112.07938

    2022 arXiv

  6. [14]

    Efficient and light-weight federated learning via asynchronous dis- tributed dropout,

    C. Dun, M. Hipolito, C. Jermaine, D. Dimitriadis, and A. Kyrillidis, “Efficient and light-weight federated learning via asynchronous dis- tributed dropout,” inInternational Conference on Artificial Intelligence and Statistics. PMLR, 2023, pp. 6630–6660

    2023

  7. [15]

    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. Cummingset al., “Advances and open problems in federated learning,”Foundations and trends® in machine learning, vol. 14, no. 1–2, pp. 1–210, 2021

    2021

  8. [16]

    FedSZ: Leveraging error-bounded lossy compression for federated learning communications,

    G. Wilkins, S. Di, J. C. Calhoun, Z. Li, K. Kim, R. Underwood, R. Mortier, and F. Cappello, “FedSZ: Leveraging error-bounded lossy compression for federated learning communications,” in2024 IEEE 44th International Conference on Distributed Computing Systems (ICDCS). IEEE, 2024...

    2024

  9. [17]

    FedSpaLLM: Federated pruning of large language models,

    G. Bai, Y . Li, Z. Li, L. Zhao, and K. Kim, “FedSpaLLM: Federated pruning of large language models,”arXiv preprint arXiv:2410.14852, 2024

    2024 arXiv

  10. [18]

    Fed-SB: A silver bullet for extreme communication efficiency and performance in (private) federated lora fine-tuning,

    R. Singhal, K. Ponkshe, R. Vartak, L. R. Varshney, and P. Vepakomma, “Fed-SB: A silver bullet for extreme communication efficiency and performance in (private) federated lora fine-tuning,”arXiv preprint arXiv:2502.15436, 2025

    2025

  11. [19]

    SpotDNN: Provisioning spot instances for predictable distributed DNN training in the cloud,

    R. Shang, F. Xu, Z. Bai, L. Chen, Z. Zhou, and F. Liu, “SpotDNN: Provisioning spot instances for predictable distributed DNN training in the cloud,” in2023 IEEE/ACM 31st International Symposium on Quality of Service (IWQoS). IEEE, 2023, pp. 1–10

    2023

  12. [20]

    Spotnik: Designing distributed machine learning for transient cloud resources,

    M. Wagenl ¨ander, L. Mai, G. Li, and P. Pietzuch, “Spotnik: Designing distributed machine learning for transient cloud resources,” in12th USENIX Workshop on Hot Topics in Cloud Computing (HotCloud 20), 2020

    2020

  13. [21]

    Cynthia: Cost-efficient cloud resource provisioning for predictable distributed deep neural network training,

    H. Zheng, F. Xu, L. Chen, Z. Zhou, and F. Liu, “Cynthia: Cost-efficient cloud resource provisioning for predictable distributed deep neural network training,” inProceedings of the 48th International Conference on Parallel Processing, 2019, pp. 1–11

    2019

  14. [22]

    Convex optimization of markov decision processes based on z transform: A theoretical framework for two-space decomposition and linear programming recon- struction,

    S. Qiu, H. Wang, Y . Zhang, Z. Ke, and Z. Li, “Convex optimization of markov decision processes based on z transform: A theoretical framework for two-space decomposition and linear programming recon- struction,”Mathematics, vol. 13, no. 11, p. 1765, 2025

    2025

  15. [23]

    Oort: Efficient federated learning via guided participant selection,

    F. Lai, X. Zhu, H. V . Madhyastha, and M. Chowdhury, “Oort: Efficient federated learning via guided participant selection,” in15th{USENIX} Symposium on Operating Systems Design and Implementation ({OSDI} 21), 2021, pp. 19–35

    2021

  16. [24]

    Haccs: Heterogeneity-aware clustered client selection for accelerated federated learning,

    J. Wolfrath, N. Sreekumar, D. Kumar, Y . Wang, and A. Chandra, “Haccs: Heterogeneity-aware clustered client selection for accelerated federated learning,” in2022 IEEE international parallel and distributed processing symposium (IPDPS). IEEE, 2022, pp. 985–995

    2022

  17. [25]

    HeteroFL: Computation and com- munication efficient federated learning for heterogeneous clients,

    E. Diao, J. Ding, and V . Tarokh, “HeteroFL: Computation and com- munication efficient federated learning for heterogeneous clients,”arXiv preprint arXiv:2010.01264, 2020

    2010 arXiv

  18. [26]

    Stragglers-aware low-latency synchronous federated learning via layer-wise model updates,

    N. Lang, A. Cohen, and N. Shlezinger, “Stragglers-aware low-latency synchronous federated learning via layer-wise model updates,”IEEE Transactions on Communications, 2024

    2024

  19. [27]

    Ray: A distributed framework for emerging AI applications,

    P. Moritz, R. Nishihara, S. Wang, A. Tumanov, R. Liaw, E. Liang, M. Elibol, Z. Yang, W. Paul, M. I. Jordan, and I. Stoica, “Ray: A distributed framework for emerging AI applications,” 2018. [Online]. Available: https://arxiv.org/abs/1712.05889

    2018 arXiv

  20. [28]

    Cloud-bursting and autoscaling for python-native scientific workflows using ray,

    T. Liu, M. Ellis, C. Costa, C. Misale, S. Kokkila-Schumacher, J. Jung, G.-J. Nam, and V . Kindratenko, “Cloud-bursting and autoscaling for python-native scientific workflows using ray,” inHigh Performance Computing: ISC High Performance 2023 International Workshops, Hamburg, G...

    2023 doi

  21. [29]

    Automated data management and learning- based scheduling for Ray-based hybrid HPC-cloud systems,

    T. Liu, H. Tao, Y . Lu, Z. Zhu, M. Ellis, S. Kokkila-Schumacher, and V . Kindratenko, “Automated data management and learning- based scheduling for Ray-based hybrid HPC-cloud systems,” in Euro-Par 2024: Parallel Processing: 30th European Conference on Parallel and Distributed ...

    2024

  22. [30]

    APPFL: open-source software framework for privacy-preserving federated learning,

    M. Ryu, Y . Kim, K. Kim, and R. K. Madduri, “APPFL: open-source software framework for privacy-preserving federated learning,” in2022 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW). IEEE, 2022, pp. 1074–1083

    2022

  23. [31]

    Advances in APPFL: A comprehensive and extensible federated learning framework,

    Z. Li, S. He, Z. Yang, M. Ryu, K. Kim, and R. Madduri, “Advances in APPFL: A comprehensive and extensible federated learning framework,” arXiv preprint arXiv:2409.11585, 2024

    2024 arXiv

  24. [32]

    The mnist database of handwritten digits,

    L. Yann, “The mnist database of handwritten digits,”R, 1998

    1998

  25. [33]

    Learning multiple layers of features from tiny images,

    A. Krizhevsky, G. Hintonet al., “Learning multiple layers of features from tiny images,” 2009

    2009

  26. [34]

    Flagship dataset of type 2 diabetes from the AI-READI project (1.0.0),

    A.-R. Consortium, “Flagship dataset of type 2 diabetes from the AI-READI project (1.0.0),” 2024. [Online]. Available: https: //doi.org/10.60775/fairhub.1

    2024 doi

  27. [35]

    Flamby: Datasets and benchmarks for cross-silo federated learning in realistic healthcare settings,

    J. Ogier du Terrail, S.-S. Ayed, E. Cyffers, F. Grimberg, C. He, R. Loeb, P. Mangold, T. Marchand, O. Marfoq, E. Mushtaqet al., “Flamby: Datasets and benchmarks for cross-silo federated learning in realistic healthcare settings,”Advances in Neural Information Processing System...

    2022

  28. [36]

    Efficientnet: Rethinking model scaling for con- volutional neural networks,

    M. Tan and Q. Le, “Efficientnet: Rethinking model scaling for con- volutional neural networks,” inInternational conference on machine learning. PMLR, 2019, pp. 6105–6114

    2019

  29. [37]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778

    2016

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.