Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

Denoising and Adaptive Online Vertical Federated Learning for Sequential Multi-Sensor Data in Industrial Internet of Things

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

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

Pith's one-line read Denoising plus adaptive local steps tightens the regret bound for online vertical federated learning in noisy industrial settings.

desk verdict A plausible new combination of online VFL, denoising, and DRL-based local iterations, but the proof is missing and the denoising mechanism assumes the very clean data the paper says isn't available. read the letter →

arxiv 2501.01693 v1 pith:PG63U4G3 submitted 2025-01-03 cs.LG cs.NI

Heqiang Wang , Xiaoxiong Zhong , Kang Liu , Fangming Liu , Weizhe Zhang This is my paper · ORCID
classification cs.LGcs.NI MSC 68T0568W27
keywords IndustrialInternetofThingsVerticalFederatedLearningOnlineDeepReinforcementdenoisingautoencoderregretboundadaptivelocaliterationsmulti-sensordata
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

DAO-VFL targets an industrial assembly line where multiple sensors with distinct feature spaces train a shared model online by uploading feature embeddings to a server over noisy wireless links. The paper's central claim is that a server-side denoising autoencoder plus per-sensor adaptive local iteration counts makes this online problem learnable: under convexity and bounded-gradient assumptions, cumulative regret grows as $O(\sqrt{T}+T\beta_d)$, where $\beta_d$ measures how close the denoised gradient is to the clean gradient. Because $\beta_d$ sits inside the regret bound, effective denoising directly buys a tighter guarantee, and keeping local iteration counts similar across sensors tightens the bound as well. If the claim holds, vertical federated learning can run on streaming industrial data despite transmission noise and heterogeneous sensor capabilities.

What carries the argument

The load-bearing mechanism is the server-side denoising autoencoder applied to noisy feature embeddings, paired with a deep reinforcement learning agent that outputs the local iteration count $E_{t,k}$ for each sensor in every global round. The denoising autoencoder is trained during an initial denoising learning period to map the noisy embedding $\tilde{h}_k(\theta_k;\mathbf{x}_k)$ to a denoised $\hat{h}_k(\theta_k;\mathbf{x}_k)$, which is what makes the gradient-deviation bound $\beta_d$ in Assumption 5 small. The server bundles the head model and all denoised embeddings into the model representation $\hat{\Phi}^{t,0}$, distributes it to sensors, and each sensor runs online gradient descent for $E_{t,k}$ iterations. The regret analysis then separates the denoising error $\beta_d$ from the iteration-count terms $E_{\max}$ and $E_{\min}$, so the theorem's tightness depends on exactly the two components the algorithm controls.

What would settle it

Run DAO-VFL in a setting where no clean feature embeddings ever reach the server, so the autoencoder can only be trained on noisy pairs, and compare cumulative regret against the paper's noise-included and noise-excluded baselines; if the denoised run does not beat the noise-included run, or if empirical regret grows faster than $O(\sqrt{T}+T\beta_d)$, the central claim is falsified.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that online vertical federated learning remains tractable in noisy, heterogeneous industrial settings when the server denoises incoming feature embeddings and each sensor's local iteration count is chosen adaptively. Theorem 1 gives, under Assumptions 1 to 5, the regret bound $$\mathrm{Reg}_T \le \frac{\|\$Theta^{{1,0}}$-\Theta^*\|^2}{2\eta E_{\min}} + \frac{\eta T D \$beta_d^{2}$}{E_{\min}} + \frac{\eta T E_{\max} $L^{2}$ K}{E_{\min}} + 2DT\rho(\eta\$\lambda$ E_{\max}L + \beta_d).$$ With $\eta = O(1/\sqrt{T})$, this is $O(\sqrt{T}+T\beta_d)$, where $\beta_d$ is the per-coordinate bound on the difference between denoised and clean gradients from Assumption 5. The paper argues that replacing $\beta_d$ with the larger noise-level bound $\beta_n$ exposes why denoising helps, and that small $E_{\max}$ with large $E_{\min}$ tightens the bound. Experiments on CIFAR-10 and C-MAPSS are presented as evidence that the denoised variant matches or improves on the noise-free baseline, and that the DRL-based iteration choices reduce total latency and improve the reward objective.

Load-bearing premise

Section IV.2 assumes that during the first $T_{dl}$ global rounds the server receives clean, noise-free feature embeddings to train the denoising autoencoder, and Remark 2 concedes that clean data is typically required; since the paper's motivating scenario says all sensor-to-server wireless transmissions are noisy, this clean-access period is the load-bearing premise.

Editorial extensions

If this is right

  • A practitioner can tune the denoiser and the local iteration schedule by targeting $\beta_d$, $E_{\max}$, and $E_{\min}$ instead of tuning the whole system empirically.
  • Partially effective denoising still improves the guarantee, because any $\beta_d$ smaller than the raw noise bound $\beta_n$ makes the regret bound tighter.
  • Keeping per-sensor local iteration counts close to each other is not only a latency fairness goal; the regret bound makes it a formal requirement.
  • The algorithm can be applied to streaming IIoT data without waiting for a static dataset, since both feature and head models update by online gradient steps.
  • The reward design of the DRL problem gives a concrete trade-off among accuracy, total latency, and iteration disparity, and the learned policy executes that trade-off each round.

Reading between the lines

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

  • The paper leaves implicit that the clean-embedding training period could be replaced by self-supervised denoising trained only on noisy pairs; a testable extension is to analyze whether the regret bound still holds with $\beta_d$ defined against such a denoiser.
  • Because the regret term $O(T\beta_d)$ is additive, there is a point of diminishing returns where further denoising effort stops mattering; measuring $\beta_d$ empirically would let practitioners stop investing in denoising once other terms dominate.
  • The adaptive iteration mechanism is effectively a straggler-mitigation policy: sensors with slow CPUs or poor channels receive fewer local iterations, which connects DAO-VFL to asynchronous and heterogeneous federated learning beyond the assembly-line setting.
  • A direct experimental check would measure the empirical $\beta_d$ of the trained autoencoder and compare the observed regret against the bound's prediction, turning Theorem 1 into a deployable diagnostic.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The manuscript studies online vertical federated learning for multi-sensor industrial assembly lines. It proposes DAO-VFL, in which sensors upload low-dimensional feature embeddings to a server; a denoising autoencoder (DAE) mitigates communication noise, and a deep-reinforcement-learning controller selects per-sensor local iteration counts. The paper derives a regret bound (Theorem 1) under convexity and Lipschitz assumptions, states that effective denoising tightens the bound, and reports experiments on CIFAR-10 and C-MAPSS against noise-included/excluded and homogeneous/heterogeneous baselines. The adaptive-iteration results show latency and reward benefits, and the denoising results show improved accuracy/RMSE relative to the noisy baseline.

Significance. If the theoretical claims were established, the paper would contribute to online VFL in practical IIoT settings by jointly addressing noise and sensor heterogeneity. The problem formulation is relevant and the experimental study is fairly broad. However, the central regret theorem is unverifiable because the proof is not present in the manuscript, its assumptions do not hold in the experiments, and the key denoising benefit is encoded in an assumption rather than derived. The clean-data requirement during the denoising learning period is inconsistent with the paper's own noisy-channel premise. These issues undermine the paper's main claims as submitted. I also note that the manuscript does not ship code or machine-checked proofs, so the theory must stand on the text alone.

major comments (4)
  1. [Section V, Theorem 1 and proof] The proof of Theorem 1 is deferred to an appendix that is not present in the manuscript; the only indication is the line 'The proof can be found in Appendix.' As a result, the regret bound cannot be independently checked, and the central theoretical contribution is not supported by the submission. A complete proof must be included.
  2. [Section V, Assumption 1 vs Section VII-C] Theorem 1 relies on Assumption 1, which requires Ft to be convex in Θ. The experiments in Section VII-C use deep convolutional feature extractors (13 conv layers for CIFAR-10 and 2 conv layers for C-MAPSS) followed by nonlinear heads, which are not convex. Therefore the regret guarantee cannot be invoked for the reported empirical results, and the claimed theoretical basis for the observed improvements is absent.
  3. [Section IV.2, Eq. (3) and Remark 2] The DAE training objective in Eq. (3) requires paired clean and noisy feature embeddings, and Section IV.2 assumes that 'the original feature embeddings from the sensors are available to the server' during the first Tdl rounds. This contradicts the paper's premise that wireless transmission corrupts all sensor-to-server embeddings. If the channel is noisy for every upload, no clean reference exists; if it is clean for Tdl, the noise model is undefined. Remark 2 acknowledges the issue and mentions Noise2Noise/Noise2Void but states they are not explored. Thus the central noise-reduction mechanism is not implementable under the stated system model.
  4. [Section V, Assumption 5 and subsequent findings] Assumption 5 defines βd as the elementwise bound between the denoised gradient and the original gradient. Since the analysis never derives βd from the DAE training in Eq. (3), from the quantization/noise model in Section VII-E, or from any estimation procedure, the statement that 'effective denoising leads to a tighter regret bound' is essentially an assumption, not a conclusion. The comparison with βn is also made without formally defining βn. To support the paper's central claim, βd must be derived or measured under a concrete noise model, with a demonstrated reduction relative to βn.
minor comments (5)
  1. [Algorithm 2, line 4] The tuple lists Υco_t,k twice; the second entry should be the communication latency Υcm_t,k.
  2. [Eq. (20) and Algorithm 2] Eq. (20) is called a 'gain function' but it is a squared temporal-difference loss to be minimized by the critic; the description of updating the critic by 'maximizing the reward via Eq. 20' is confusing.
  3. [Section VII-F] The benchmark 'DAO-NR' appears in figures and text (e.g., Fig. 6) without being defined; the reader must infer that it is the noise-reduction variant of DAO-VFL.
  4. [Theorem 1 and Eq. (2)] Theorem 1's left-hand side uses E_t[Ft(...)] while Eq. (2) defines regret without an expectation; the meaning of E_t is not explained.
  5. [Section IV.2 and Section VII-D] The parameters µ0 and µ in the collection latency definition, Eq. (10), are not defined before use; they are only assigned example values in Section VII-D.

Circularity Check

3 steps flagged · score 6.0 of 10

The claimed noise-reduction benefit is largely built into Assumption 5, and the DAE training process assumes clean embeddings that the noise model denies; the adaptive-iteration experiments also optimize a reward containing test accuracy, making part of the empirical support circular.

  1. self definitional [Section V, Assumption 5 and the discussion after Theorem 1 (Eq. 9)]
    "Assumption 5. The arbitrary vector element d in the overall gradient, adjusted through a denoising method, has a bounded range range of variation as: |Ĝ t,τ k,d − G t,τ k,d | ≤ βd. ... It becomes clear that the primary factors influencing the regret bound are the magnitudes of βn and βd. If the noise reduction method is effective, it consistently leads to a tighter regret bound."

    beta_d is defined as the bound on the deviation between the denoised gradient and the original gradient. The regret bound's beta_d term therefore restates Assumption 5 rather than deriving the effect of the DAE. 'Effective noise reduction' is by definition equivalent to a small beta_d, so the claimed conclusion that denoising tightens the bound is a tautology unless the paper separately proves that the DAE training yields beta_d < beta_n. No such derivation is provided; the proof of Theorem 1 is deferred to a missing appendix.

  2. other [Section IV-2 (Feature Embedding Denoising) and Eq. (3)]
    "During the initial Tdl global rounds, also referred to as the denoising learning period, it is assumed that the original feature embeddings from the sensors are available to the server. ... arg min θd E{˜ht k(θt,0 k ), ht k(θt,0 k )} { Ls(Λθd (˜ht k(θt,0 k )), ht k(θt,0 k )) }"

    The system model states that feature embeddings transmitted through the industrial wireless network are 'inevitably affected by noise,' yet the DAE training target in Eq. (3) requires the clean original embeddings ht k. The clean-data assumption supplies the very signal the denoising mechanism is supposed to recover. Under the stated noise model, the server cannot obtain these clean originals without an unmodeled side channel. Consequently, the claimed reduction of beta_d and the resulting tighter regret bound are not established; the positive denoising result in experiments is an artifact of assuming away the paper's own central noise premise.

1 more flagged steps
  1. fitted input called prediction [Section VI, Eq. (19) and Algorithm 2, step 9]
    "Rt = α1Acc(t) − α2Υt − α3Ht ... Infer on test dataset to obtain test accuracy. ... The server calculate the reward Rt."

    The DRL policy for adaptive local iteration decisions is trained to maximize a reward that directly includes test accuracy Acc(t), and Algorithm 2 explicitly obtains Acc(t) from the test dataset. The paper then presents DAO-PPO's test accuracy and test loss as evidence that adaptive decisions improve learning performance. Since the policy is fitted to the evaluation metric, the experimental 'superior performance' is partly forced by the training objective rather than being an independent prediction of the regret analysis or a fair comparison against fixed baselines.

full rationale

DAO-VFL is not wholly circular: the online OGD regret framework, the convexity/Lipschitz assumptions, and the Emax/Emin dependence in Theorem 1 are standard conditional results, and the experimental comparison against CIFAR-10 and C-MAPSS is externally grounded. However, the paper's headline claim that its denoising mechanism yields a tighter regret bound reduces to Assumption 5: beta_d is defined as the deviation between denoised and original gradients, so saying 'effective denoising gives a tighter bound' is a restatement of the assumption, not a derivation from the DAE. The DAE training further assumes clean original embeddings are available during Tdl, contradicting the model's premise that all uploaded embeddings are noisy; without clean data, Eq. (3) is not implementable and the experimental denoising gain is not supported by the stated system model. Additionally, the adaptive-iteration experiments are weakened because the DRL reward includes test accuracy, so the policy is directly optimizing the reported evaluation metric. The missing appendix for Theorem 1 prevents independent verification of the bound, but the more fundamental circularity is the beta_d assumption and the clean-data training premise. Overall, the central noise-reduction prediction is substantially by construction, while other components retain independent content; hence a score of 6 is appropriate.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central claims rest on several assumptions that are either unverified or conflict with the experimental setup. Most importantly, the DAE requires clean embeddings during T_dl while the motivating scenario assumes noisy communication, and the regret theorem assumes convexity while the experiments use deep nonlinear models. The alpha weights, T_dl, and E_max are hand-chosen parameters that affect the reported results but are not fully specified.

free parameters (3)
  • alpha1, alpha2, alpha3 = not reported
    Weights in the optimization problem P1 (Eq. 16) and the reward (Eq. 19). They control the trade-off between accuracy, latency, and local iteration disparity, but their values are not given. All reported latency and reward comparisons depend on these unspecified weights.
  • T_dl (denoising learning period) = 40 in the main noise reduction experiments
    Hand-chosen hyperparameter controlling how many global rounds the DAE is trained. Figure 7 shows performance improves with T_dl, so the reported results depend on this tuning choice.
  • E_max (maximum local iterations) = not reported
    Empirically predefined upper limit for local iterations per sensor per round, used both in the action space and in the HO benchmark. Its value is not stated in the experimental setup.
assumptions (6)
  • domain assumption Assumption 1: The per-round loss function F_t is convex and differentiable with respect to the global model.
    Section V. Standard for online convex optimization, but it does not hold for the deep convolutional feature extractors and nonlinear head used in the CIFAR-10 and C-MAPSS experiments, so the regret theorem does not cover the evaluated setting.
  • ad hoc to paper Assumption 5: The denoised gradient is elementwise within beta_d of the clean gradient.
    Section V. This assumption effectively assumes the noise-reduction benefit: the regret bound tightens as beta_d decreases, but the paper does not prove the DAE achieves beta_d smaller than the noisy gradient bound beta_n.
  • ad hoc to paper Clean feature embeddings are available to the server during the initial T_dl global rounds.
    Section IV.2. This is required to train the denoising autoencoder with paired noisy and clean embeddings, but it contradicts the paper's premise that wireless communication is always noisy in the industrial environment. No practical mechanism is given for obtaining clean embeddings during online operation.
  • domain assumption All sensors and the server have access to the label information.
    Remark 1. The paper assumes a low-risk environment where label sharing is acceptable, which is a domain-specific simplification.
  • domain assumption Server-to-sensor downlink is noise-free because the server uses directional antennas.
    Section IV.3. The model representation is distributed without noise based on this assumption, while the sensor-to-server uplink is treated as noisy. This asymmetry is plausible but not modeled or validated.
  • standard math Assumptions 2, 3, and 4: bounded gradients, a Lipschitz-like gradient variation condition, and bounded model parameters.
    Section V. Standard boundedness conditions used in online convex optimization regret analysis. They are stated but not verified for the experimental models.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Denoising and Adaptive Online Vertical Federated Learning for Sequential Multi-Sensor Data in Industrial Internet of Things." pith.science (2026). https://pith.science/paper/PG63U4G3

@misc{pith2026250101693,
  author       = {Pith},
  title        = {Pith review of: Denoising and Adaptive Online Vertical Federated Learning for Sequential Multi-Sensor Data in Industrial Internet of Things},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PG63U4G3}},
  note         = {Machine review of arXiv:2501.01693}
}
read the original abstract

With the continuous improvement in the computational capabilities of edge devices such as intelligent sensors in the Industrial Internet of Things, these sensors are no longer limited to mere data collection but are increasingly capable of performing complex computational tasks. This advancement provides both the motivation and the foundation for adopting distributed learning approaches. This study focuses on an industrial assembly line scenario where multiple sensors, distributed across various locations, sequentially collect real-time data characterized by distinct feature spaces. To leverage the computational potential of these sensors while addressing the challenges of communication overhead and privacy concerns inherent in centralized learning, we propose the Denoising and Adaptive Online Vertical Federated Learning (DAO-VFL) algorithm. Tailored to the industrial assembly line scenario, DAO-VFL effectively manages continuous data streams and adapts to shifting learning objectives. Furthermore, it can address critical challenges prevalent in industrial environment, such as communication noise and heterogeneity of sensor capabilities. To support the proposed algorithm, we provide a comprehensive theoretical analysis, highlighting the effects of noise reduction and adaptive local iteration decisions on the regret bound. Experimental results on two real-world datasets further demonstrate the superior performance of DAO-VFL compared to benchmarks algorithms.

Figures

Figures reproduced from arXiv: 2501.01693 by the authors.

Figure 1
Figure 1. DAO-VFL for IIoT-Based Assembly Line The IIoT encompasses a broad array of application sce￾narios, each presenting its own set of problems, challenges, and potential solutions. In this work, we focus on an indus￾trial assembly line scenario, as illustrated in [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Challenges and Solutions in DAO-VFL based on the real-time status of the sensors in the current global round. The approach for deciding the number of local iterations for each sensor will be discussed in the later section. Each local training iteration is indexed as τ = 0, 1, 2, ..., Et,k. Notably, the DAO-VFL algorithm is designed for an online synchronized scenario, even though sensors perform varying numbers of l… view at source ↗
Figure 3
Figure 3. The Architecture of DAO-VFL Noise Reduction [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 5
Figure 5. Figure 5: Timeline of Sensors in One Global Round For simplicity in the analysis, we assume that sensors collect data sequentially based on their index, such that sensors with smaller indices collect data earlier. 2) Communication Latency: In this scenario, sensors share bandwid…
Figure 6
Figure 6. Figure 6: Performance Comparison of DAO-VFL and Bench [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Performance Comparison of DAO-VFL and Bench [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 9
Figure 9. Figure 9: Comparison of Total Latency. Comparison of Reward. In this section, we compare the reward of DAO-PPO with benchmarks within a single run [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: Comparison of Reward. The experimental results above demonstrate the superiority of the proposed DAO-VFL algorithm over its corresponding benchmarks in both noise reduction and adaptive local iteration decision-making. VIII. CONCLUSION In this work, we proposed the DA…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Multimodal Online Federated Learning with Modality Missing in Internet of Things

    cs.LG 2025-05 conditional novelty 5.0 of 10

    Introduces MMO-FL, an online federated multimodal learning framework with a prototype-based algorithm, PMM, for compensating missing sensor modalities.

  2. MTF-Grasp: A Multi-tier Federated Learning Approach for Robotic Grasping

    cs.LG 2025-07 reject novelty 4.0 of 10

    MTF-Grasp, a two-tier federated learning method that seeds low-data robots with models pre-trained by high-quality clients, reports up to 8% higher grasp accuracy than vanilla FedAvg under data quantity skew.

Reference graph

Works this paper leans on

48 extracted references · 38 canonical work pages · cited by 2 Pith papers

  1. [38]

    Online vertical federated learning for cooper- ative spectrum sensing,

    H. Wang and J. Xu, “Online vertical federated learning for cooper- ative spectrum sensing,” IEEE Transactions on Cognitive Communi- cations and Networking , 2024

    work page 2024

  2. [1]

    The industrial internet of things (iiot): An analysis framework,

    H. Boyes, B. Hallaq, J. Cunningham, and T. Watson, “The industrial internet of things (iiot): An analysis framework,” Computers in Industry, vol. 101, pp. 1–12, 2018

    work page 2018

  3. [2]

    Industry 4.0: A survey on technologies, applications and open research issues,

    Y . Lu, “Industry 4.0: A survey on technologies, applications and open research issues,” Journal of Industrial Information Integration, vol. 6, pp. 1–10, 2017

    work page 2017

  4. [3]

    Towards fully au- tonomous driving: Systems and algorithms,

    J. Levinson, J. Askeland, J. Becker, J. Dolson, D. Held, S. Kammel, J. Z. Kolter, D. Langer, O. Pink, V . Pratt et al. , “Towards fully au- tonomous driving: Systems and algorithms,” in 2011 IEEE Intelligent Vehicles Symposium (IV). IEEE, 2011, pp. 163–168

    work page 2011

  5. [4]

    Context-sensitive access in indus- trial internet of things (iiot) healthcare applications,

    F. Al-Turjman and S. Alturjman, “Context-sensitive access in indus- trial internet of things (iiot) healthcare applications,” IEEE Transac- tions on Industrial Informatics , vol. 14, no. 6, pp. 2736–2744, 2018

    work page 2018

  6. [5]

    Federated learning in mobile edge networks: A comprehensive survey,

    W. Y . B. Lim, N. C. Luong, D. T. Hoang, Y . Jiao, Y .-C. Liang, Q. Yang, D. Niyato, and C. Miao, “Federated learning in mobile edge networks: A comprehensive survey,” IEEE Communications Surveys and Tutorials, vol. 22, no. 3, pp. 2031–2063, 2020

    work page 2020

  7. [6]

    Vertical federated learning: Concepts, advances, and challenges,

    Y . Liu, Y . Kang, T. Zou, Y . Pu, Y . He, X. Ye, Y . Ouyang, Y .-Q. Zhang, and Q. Yang, “Vertical federated learning: Concepts, advances, and challenges,” IEEE Transactions on Knowledge and Data Engineering, 2024

    work page 2024

  8. [7]

    Vertical federated learning: challenges, methodologies and experiments,

    K. Wei, J. Li, C. Ma, M. Ding, S. Wei, F. Wu, G. Chen, and T. Ran- baduge, “Vertical federated learning: challenges, methodologies and experiments,” arXiv preprint arXiv:2202.04309 , 2022

    arXiv 2022

Show all 48 references
  1. [8]

    Noise removal in the presence of significant anomalies for industrial iot sensor data in manufacturing,

    Y . Liu, T. Dillon, W. Yu, W. Rahayu, and F. Mostafa, “Noise removal in the presence of significant anomalies for industrial iot sensor data in manufacturing,” IEEE Internet of Things Journal , vol. 7, no. 8, pp. 7084–7096, 2020

    2020

  2. [9]

    Online optimization of wireless powered mobile-edge computing for heterogeneous industrial internet of things,

    H. Wu, X. Lyu, and H. Tian, “Online optimization of wireless powered mobile-edge computing for heterogeneous industrial internet of things,” IEEE Internet of Things Journal , vol. 6, no. 6, pp. 9880– 9892, 2019

    2019

  3. [10]

    In- dustrial internet of things: Challenges, opportunities, and directions,

    E. Sisinni, A. Saifullah, S. Han, U. Jennehag, and M. Gidlund, “In- dustrial internet of things: Challenges, opportunities, and directions,” IEEE Transactions on Industrial Informatics , vol. 14, no. 11, pp. 4724–4734, 2018

    2018

  4. [11]

    Federated learning for industrial internet of things in future industries,

    D. C. Nguyen, M. Ding, P. N. Pathirana, A. Seneviratne, J. Li, D. Niy- ato, and H. V . Poor, “Federated learning for industrial internet of things in future industries,” IEEE Wireless Communications, vol. 28, no. 6, pp. 192–199, 2021

    2021

  5. [12]

    Fusion of feder- ated learning and industrial internet of things: A survey,

    P. Boobalan, S. P. Ramu, Q.-V . Pham, K. Dev, S. Pandya, P. K. R. Maddikunta, T. R. Gadekallu, and T. Huynh-The, “Fusion of feder- ated learning and industrial internet of things: A survey,” Computer Networks, vol. 212, p. 109048, 2022

    2022

  6. [13]

    Federated transfer learning based cross-domain prediction for smart manufac- turing,

    I. Kevin, K. Wang, X. Zhou, W. Liang, Z. Yan, and J. She, “Federated transfer learning based cross-domain prediction for smart manufac- turing,” IEEE Transactions on Industrial Informatics , vol. 18, no. 6, pp. 4088–4096, 2021

    2021

  7. [14]

    An efficient and reliable asynchronous federated learning scheme for smart public transportation,

    C. Xu, Y . Qu, T. H. Luan, P. W. Eklund, Y . Xiang, and L. Gao, “An efficient and reliable asynchronous federated learning scheme for smart public transportation,” IEEE Transactions on Vehicular Technology, vol. 72, no. 5, pp. 6584–6598, 2022

    2022

  8. [15]

    Secure and efficient federated learning for smart grid with edge- cloud collaboration,

    Z. Su, Y . Wang, T. H. Luan, N. Zhang, F. Li, T. Chen, and H. Cao, “Secure and efficient federated learning for smart grid with edge- cloud collaboration,” IEEE Transactions on Industrial Informatics , vol. 18, no. 2, pp. 1333–1344, 2021

    2021

  9. [16]

    Toward accurate anomaly detection in industrial internet of things using hierarchical federated learning,

    X. Wang, S. Garg, H. Lin, J. Hu, G. Kaddoum, M. J. Piran, and M. S. Hossain, “Toward accurate anomaly detection in industrial internet of things using hierarchical federated learning,” IEEE Internet of Things Journal, vol. 9, no. 10, pp. 7110–7119, 2021. 13

    2021

  10. [17]

    Towards ever- evolution network threats: a hierarchical federated class-incremental learning approach for network intrusion detection in iiot,

    J. Mao, Z. Wei, B. Li, R. Zhang, and L. Song, “Towards ever- evolution network threats: a hierarchical federated class-incremental learning approach for network intrusion detection in iiot,” IEEE Internet of Things Journal , 2024

    2024

  11. [18]

    Federated learning for distributed iiot intrusion detection using transfer approaches,

    J. Zhang, C. Luo, M. Carpenter, and G. Min, “Federated learning for distributed iiot intrusion detection using transfer approaches,” IEEE Transactions on Industrial Informatics, vol. 19, no. 7, pp. 8159–8169, 2022

    2022

  12. [19]

    On the local cache update rules in streaming federated learning,

    H. Wang, J. Bian, and J. Xu, “On the local cache update rules in streaming federated learning,” IEEE Internet of Things Journal, 2023

    2023

  13. [20]

    Private federated learning on vertically partitioned data via entity resolution and additively homomorphic encryption,

    S. Hardy, W. Henecka, H. Ivey-Law, R. Nock, G. Patrini, G. Smith, and B. Thorne, “Private federated learning on vertically partitioned data via entity resolution and additively homomorphic encryption,” CoRR, vol. abs/1711.10677, 2017

    2017 arXiv

  14. [21]

    From federated learning to federated neural architecture search: a survey,

    H. Zhu, H. Zhang, and Y . Jin, “From federated learning to federated neural architecture search: a survey,” Complex & Intelligent Systems , vol. 7, no. 2, pp. 639–657, 2021

    2021

  15. [22]

    Vertical federated learning-based feature selection with non-overlapping sample utilization,

    S. Feng, “Vertical federated learning-based feature selection with non-overlapping sample utilization,” Expert Systems with Applica- tions, vol. 208, p. 118097, 2022

    2022

  16. [23]

    Fedcvt: semi-supervised vertical federated learning with cross-view training,

    Y . Kang, Y . Liu, and X. Liang, “Fedcvt: semi-supervised vertical federated learning with cross-view training,” ACM Transactions on Intelligent Systems and Technology (TIST) , vol. 13, no. 4, pp. 1–16, 2022

    2022

  17. [24]

    Fair and efficient contribution valuation for vertical federated learn- ing,

    Z. Fan, H. Fang, Z. Zhou, J. Pei, M. P. Friedlander, and Y . Zhang, “Fair and efficient contribution valuation for vertical federated learn- ing,” arXiv preprint arXiv:2201.02658 , 2022

    2022 arXiv

  18. [25]

    Label leakage and protec- tion from forward embedding in vertical federated learning,

    J. Sun, X. Yang, Y . Yao, and C. Wang, “Label leakage and protec- tion from forward embedding in vertical federated learning,” arXiv preprint arXiv:2203.01451, 2022

    2022 arXiv

  19. [26]

    Defending against reconstruction attack in vertical federated learning,

    J. Sun, Y . Yao, W. Gao, J. Xie, and C. Wang, “Defending against reconstruction attack in vertical federated learning,” arXiv preprint arXiv:2107.09898, 2021

    2021 arXiv

  20. [27]

    Flexible vertical federated learning with heterogeneous parties,

    T. Castiglia, S. Wang, and S. Patterson, “Flexible vertical federated learning with heterogeneous parties,” IEEE Transactions on Neural Networks and Learning Systems , 2023

    2023

  21. [28]

    Adaptive vertical federated learning on unbalanced fea- tures,

    J. Zhang, S. Guo, Z. Qu, D. Zeng, H. Wang, Q. Liu, and A. Y . Zomaya, “Adaptive vertical federated learning on unbalanced fea- tures,” IEEE Transactions on Parallel and Distributed Systems , vol. 33, no. 12, pp. 4006–4018, 2022

    2022

  22. [29]

    Compressed-vfl: Communication-efficient learning with vertically partitioned data,

    T. J. Castiglia, A. Das, S. Wang, and S. Patterson, “Compressed-vfl: Communication-efficient learning with vertically partitioned data,” in International Conference on Machine Learning . PMLR, 2022, pp. 2738–2766

    2022

  23. [30]

    Computation and communication efficient lightweighting vertical federated learning,

    H. Wang, J. Bian, and L. Wang, “Computation and communication efficient lightweighting vertical federated learning,” arXiv preprint arXiv:2404.00466, 2024

    2024 arXiv

  24. [31]

    Exploiting data sparsity in secure cross-platform social recommendation,

    J. Cui, C. Chen, L. Lyu, C. Yang, and W. Li, “Exploiting data sparsity in secure cross-platform social recommendation,” Advances in Neural Information Processing Systems , vol. 34, pp. 10 524–10 534, 2021

    2021

  25. [32]

    A homomorphic-encryption-based vertical federated learning scheme for rick management,

    W. Ou, J. Zeng, Z. Guo, W. Yan, D. Liu, and S. Fuentes, “A homomorphic-encryption-based vertical federated learning scheme for rick management,” Computer Science and Information Systems , vol. 17, no. 3, pp. 819–834, 2020

    2020

  26. [33]

    Vafl: a method of vertical asynchronous federated learning,

    T. Chen, X. Jin, Y . Sun, and W. Yin, “Vafl: a method of vertical asynchronous federated learning,” arXiv preprint arXiv:2007.06081 , 2020

    2007 arXiv

  27. [34]

    Online deep learning: learning deep neural networks on the fly,

    D. Sahoo, Q. Pham, J. Lu, and S. C. Hoi, “Online deep learning: learning deep neural networks on the fly,” Proceedings of the Twenty- Seventh International Joint Conference on Artificial Intelligence (IJCAI-18), 2018

    2018

  28. [35]

    Communication-efficient randomized algo- rithm for multi-kernel online federated learning,

    S. Hong and J. Chae, “Communication-efficient randomized algo- rithm for multi-kernel online federated learning,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 44, no. 12, pp. 9872–9886, 2021

    2021

  29. [36]

    Tighter regret analysis and opti- mization of online federated learning,

    D. Kwon, J. Park, and S. Hong, “Tighter regret analysis and opti- mization of online federated learning,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2023

    2023

  30. [37]

    Online federated learning,

    A. Mitra, H. Hassani, and G. J. Pappas, “Online federated learning,” in 2021 60th IEEE Conference on Decision and Control (CDC) . IEEE, 2021, pp. 4083–4090

    2021

  31. [39]

    Industrial wireless communications over the millimeter wave spectrum: opportunities and challenges,

    M. Cheffena, “Industrial wireless communications over the millimeter wave spectrum: opportunities and challenges,”IEEE Communications Magazine, vol. 54, no. 9, pp. 66–72, 2016

    2016

  32. [40]

    Online gradient descent learning algorithms,

    Y . Ying and M. Pontil, “Online gradient descent learning algorithms,” Foundations of Computational Mathematics , vol. 8, pp. 561–596, 2008

    2008

  33. [41]

    Zurawski, Industrial communication technology handbook

    R. Zurawski, Industrial communication technology handbook . CRC Press, 2014

    2014

  34. [42]

    Noise2noise: learning image restoration without clean data,

    J. Lehtinen, J. Munkberg, J. Hasselgren, S. Laine, T. Karras, M. Ait- tala, and T. Aila, “Noise2noise: learning image restoration without clean data,” IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018

    2018

  35. [43]

    Noise2void-learning denoising from single noisy images,

    A. Krull, T.-O. Buchholz, and F. Jug, “Noise2void-learning denoising from single noisy images,” in Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition , 2019, pp. 2129– 2137

    2019

  36. [44]

    Fedqogd: Federated quantized online gradient descent with distributed time-series data,

    J. Park, D. Kwon, and S. Hong, “Fedqogd: Federated quantized online gradient descent with distributed time-series data,” in 2022 IEEE Wireless Communications and Networking Conference (WCNC) . IEEE, 2022, pp. 536–541

    2022

  37. [45]

    Deep reinforcement learning: A brief survey,

    K. Arulkumaran, M. P. Deisenroth, M. Brundage, and A. A. Bharath, “Deep reinforcement learning: A brief survey,” IEEE Signal Process- ing Magazine, vol. 34, no. 6, pp. 26–38, 2017

    2017

  38. [46]

    Actor-critic algorithms,

    V . Konda and J. Tsitsiklis, “Actor-critic algorithms,” Advances in Neural Information Processing Systems , vol. 12, 1999

    1999

  39. [47]

    Proximal policy optimization algorithms,

    J. Schulman, F. Wolski, P. Dhariwal, A. Radford, and O. Klimov, “Proximal policy optimization algorithms,” arXiv preprint arXiv:1707.06347, 2017

    2017 arXiv

  40. [48]

    Damage propaga- tion modeling for aircraft engine run-to-failure simulation,

    A. Saxena, K. Goebel, D. Simon, and N. Eklund, “Damage propaga- tion modeling for aircraft engine run-to-failure simulation,” in 2008 International Conference on Prognostics and Health Management . IEEE, 2008, pp. 1–9

    2008

Pith tools

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