FedP3E: Privacy-Preserving Prototype Exchange for Non-IID IoT Malware Detection in Cross-Silo Federated Learning

Kaushik Roy; Mahmoud Abdelsalam; Rami Darwish; Sajad Khorsandroo

REVIEW 5 major objections 6 minor 44 references

FedP3E: Privacy-Preserving Prototype Exchange for Non-IID IoT Malware Detection in Cross-Silo Federated Learning

T0 review · 5 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read FedP3E claims one round of noisy class-prototype exchange lets federated clients learn disjoint IoT malware classes, beating FedAvg and FedProx with accuracy from 95.11% to 99.57%.

desk verdict The severe non-IID result is the real contribution, but the privacy claim is asserted not demonstrated, and the numbers need seeds and error bars. read the letter →

arxiv 2507.07258 v1 pith:UGGXGXUL submitted 2025-07-09 cs.CR cs.AI

Rami Darwish , Mahmoud Abdelsalam , Sajad Khorsandroo , Kaushik Roy This is my paper · ORCID

classification cs.CRcs.AI

keywords FederatedLearningIoTMalwareDetectionCross-SiloNon-IIDDataImbalancePrototypeGaussianMixtureModelsPrivacy-Preserving

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

FedP3E is an attempt to solve a known failure of federated learning: when clients hold different, partially or fully disjoint sets of classes, global models trained by averaging local updates collapse on the classes that are rare or absent locally. The paper's claim is that a single exchange of class-wise prototypes—Gaussian mixture component means, perturbed with Gaussian noise at $\sigma=0.01$—gives every client enough statistical information about unseen classes to keep the global model accurate. On the N-BaIoT IoT traffic dataset, the authors report that this mechanism lifts accuracy from roughly 49% (where FedAvg and FedProx stall) to 95.11% in the most severe disjoint setting, and to between 99.40% and 99.57% under moderate and light non-IID conditions, with communication overhead below 10% of one full model round. If true, this gives cross-silo federated learning a cheap, one-shot way to train accurate IoT malware detectors without sharing raw data or gradients.

What carries the argument

The load-bearing object is the class-wise Gaussian Mixture prototype: the component means $\mu_j^{(c)}$ of a GMM fit to local feature vectors of class $c$, with component count selected by the Bayesian Information Criterion and then perturbed as $\tilde{\mu}_j^{(c)} = \mu_j^{(c)} + \epsilon$, $\epsilon \sim \mathcal{N}(0, \sigma^2 I)$ with $\sigma = 0.01$. This object's job is to compress one client's knowledge of a class into a few vectors, let the server merge all clients' views into global prototypes $\hat{\mu}_l^{(c)}$ via Mini-Batch K-Means, and give each client a seed set for SMOTE interpolation $x_{\mathrm{syn}} = \hat{\mu}_a^{(c)} + \lambda (\hat{\mu}_b^{(c)} - \hat{\mu}_a^{(c)})$, $\lambda \sim U(0,1)$, that produces synthetic training samples for classes the client has never seen. The whole argument that one round of prototype exchange closes the non-IID gap at under 10% of a model round's communication cost rests on this mechanism.

What would settle it

A decisive experiment is to run the severe non-IID scenario with the GMMs fit strictly on each client's 80% training split, holding the 20% test partition out of prototype construction, and then attempt a reconstruction attack on the transmitted noisy means; if accuracy collapses toward baseline or the attacker recovers the true class means with small error, the central claim fails.

Watch

Extended reading notes

Core claim

The core discovery is that indirect representation sharing through perturbed class prototypes can stand in for the missing class signal in non-IID federated learning. Each client fits a Gaussian Mixture Model to the 115-dimensional feature vectors of each class it holds, chooses the component count by Bayesian Information Criterion, and uploads only the component means after adding zero-mean Gaussian noise with standard deviation 0.01. The server clusters these noisy prototypes by class with Mini-Batch K-Means and sends the merged prototypes back to all clients, which then use SMOTE-style linear interpolation between prototype pairs to synthesize roughly 10% extra training samples for rare or absent classes. On the paper's own evaluation, this one-time, accuracy-gated exchange lets a three-client federation trained with FedAvg-style aggregation reach 99.71% accuracy in the IID case, 99.57% under light non-IID, 99.40% under moderate non-IID, and 95.11% under severe non-IID, while the baselines stay near random or collapse entirely in the worst setting. The authors present this as the reason cross-silo FL can handle statistical heterogeneity without exchanging raw data or gradients.

Load-bearing premise

The load-bearing premise is that Gaussian noise with standard deviation 0.01 added to the 115-dimensional class prototypes makes those prototypes safe to share; if that noise does not actually prevent reconstruction of sensitive class statistics, the framework's privacy guarantee collapses even if the accuracy mechanism still works.

Editorial extensions

If this is right

A system with fully disjoint client labels can still learn a global multi-class malware detector: in the severe non-IID setting, the reported accuracy is 95.11% where FedAvg and FedProx fall to roughly 49% or lower.
A single accuracy-triggered exchange is sufficient: the mechanism activates only when early global accuracy drops below 97%, so balanced settings pay almost no extra communication, and the one-time exchange itself costs less than 10% of a full model update.
The largest gains are on minority malware families: the paper reports minority-class recall improvements of about 6 percentage points over FedAvg in light non-IID and about 7 points in moderate non-IID.
The mechanism layers onto standard FedAvg aggregation and leaves the model architecture unchanged, so existing cross-silo deployments could adopt it without altering their training objective.

Reading between the lines

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

The same one-shot prototype recipe should transfer to other federated tasks with disjoint labels, such as medical imaging or financial fraud, because it only assumes that class-conditional feature vectors form clusters that a GMM can summarize.
A simpler summary than full GMM components—for example, per-class means and variances—might deliver most of the reported accuracy at lower computational cost, and comparing the two would isolate how much of the gain comes from multi-modality versus from sharing class statistics at all.
Because features are min-max scaled to [0,1], a noise scale of 0.01 is small relative to typical inter-class distances, so the natural next test is formal sensitivity analysis or a reconstruction attack on the transmitted prototypes; this would determine whether a provable privacy guarantee can be attached to the same accuracy numbers.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Quick take: FedP3E is not as novel as it claims, but it targets a real failure mode—disjoint classes across silos—and its reported 95.11% on the severe split is the kind of result that makes you stop. The mechanism is a composition of known pieces: GMM prototypes, additive noise, MiniBatch K-Means aggregation, SMOTE. The paper does not cite the FedProto line, so the novelty framing is overstated. Still, the evaluation on N-BaIoT with a three-class disjoint partition is new in the cited literature, and the communication overhead (~9.6% of one model round) is honestly quantified.

What the paper does well: the severe non-IID setup is well constructed. Each client gets one class, and FedAvg/FedProx collapse to majority-class prediction (49.39% accuracy, exactly the Mirai share). FedP3E's one-shot prototype exchange plus 10% SMOTE samples lifts that to 95.11%. That's a large, mechanistically plausible effect, and the communication cost section is clear.

Soft spots: the privacy-preserving claim is the load-bearing weakness. Section IV.D asserts that sigma=0.01 Gaussian noise on GMM means 'thwarts sample reconstruction' with no threat model, no sensitivity bound, no DP accounting, and no attack evaluation. The means are not clipped, features are min-max scaled to [0,1], and sigma=0.01 is about 1% of the feature range—averaging three prototypes per class makes the effective noise even smaller. An honest-but-curious server could estimate class centroids accurately. The title-level claim collapses unless this is fixed.

The experimental reporting is also thin: all results are single-run point estimates with no error bars or seeds. There's an internal inconsistency: Section V.C.3 says FedProx (mu=0.3) plateaus near 96.1% in moderate non-IID, but Table IX lists 94.02% for that cell. Algorithm 1 never states whether GMMs are fit on the training split only; if they're fit on full local data, the headline result could be leakage-inflated. No code is provided.

One minor point: the novelty claim would be easier to take if the paper cited and compared against existing federated prototype-sharing methods (FedProto and follow-ups), since the mechanism reduces to known components.

Overall: a serious referee could push this into a solid conditional—add a real privacy analysis or drop the privacy claim, report multiple seeds, clarify the split handling, and benchmark against prototype-sharing baselines. As it stands, the accuracy result is plausible but unverified, and the privacy property is asserted, not demonstrated.

Recommendation: I'd send it to review rather than desk reject. The severe non-IID experiment is worth refereeing, and the flaws are fixable in revision. But I would not rely on the numbers until the split ambiguity and single-run issue are resolved.

Referee Report

5 major / 6 minor

Summary. The paper proposes FedP3E, a federated learning framework for IoT malware detection in which clients periodically share Gaussian-noise-perturbed GMM component means ("prototypes") with a server; the server aggregates them via Mini-Batch K-Means, broadcasts global prototypes back to clients, and clients then use SMOTE to synthesize additional training samples for underrepresented classes. The method is evaluated on N-BaIoT with three clients under IID and three non-IID scenarios, comparing accuracy, F1, loss, training time, and communication overhead against FedAvg and FedProx. The central claims are that the one-time prototype exchange improves accuracy under non-IID conditions and that additive Gaussian noise preserves privacy.

Significance. FedP3E addresses a real problem—class imbalance and label skew in cross-silo FL—and the paper has some clear strengths: the communication-cost analysis in Section V.C.6 quantifies the one-time exchange as less than 10% of a model round, the data-partition tables are detailed, and Algorithm 1 gives a concrete procedural description. However, the manuscript's headline properties are not yet established. The privacy guarantee rests on an unexamined noise mechanism with no threat model, sensitivity bound, or differential-privacy accounting, and the empirical comparison gives FedP3E additional synthetic training data that the baselines do not receive, so the reported gains do not isolate the proposed mechanism. If the authors add a rigorous privacy analysis, error bars, and controlled ablations, the core idea could be a useful contribution to the cross-silo FL literature.

major comments (5)

[IV.D, Eq. (2)] The claim that perturbed prototypes 'thwart sample reconstruction' is not supported. The mechanism adds Gaussian noise with σ=0.01 to GMM component means of min-max-scaled [0,1] features, but no sensitivity bound is derived, no differential-privacy budget is computed, and no reconstruction or membership-inference attack is evaluated. Because the released values are means of class feature vectors and are not clipped, the sensitivity of each coordinate to a single sample is not bounded by the stated mechanism. With three prototypes per class (Section V.A.6), the effective noise is further reduced by averaging, so the asserted privacy property is not established. This is load-bearing because the title and abstract claim 'privacy-preserving'; either a formal DP guarantee with per-coordinate sensitivity analysis or an empirical attack evaluation is needed.
[V.C.4, Algorithm 1] The comparison against FedAvg and FedProx is confounded by the extra training signal given to FedP3E. In Algorithm 1 (lines 20-23), clients receive global prototypes and generate synthetic samples with SMOTE for classes missing locally, while baselines receive no such information. In the severe non-IID case (Table IX), FedP3E reaches 95.11% while all FedAvg/FedProx variants stay near 49.39% or lower, but part of this gap is attributable to the fact that FedP3E's local training set is augmented with information about classes it never observed. To support the claim that prototype exchange is the cause of the improvement, an ablation is needed—for example, FedP3E without prototype exchange, or baselines augmented with the same synthetic prototypes—so that the measured advantage is not built into the protocol.
[Table IX, V.C.3] The results contain internal inconsistencies and are single-run point estimates. Section V.C.3 states that FedProx (µ=0.3) 'plateaus near 96.1% accuracy' in the moderate non-IID scenario, but Table IX lists 94.02% for that cell. The abstract states accuracy ranges from 95.11% to 99.57%, omitting the 99.71% IID and 99.40% moderate values listed in Table IX. In addition, no seeds, confidence intervals, or repeated runs are reported anywhere, so it is impossible to assess whether the differences between methods are significant. The FedProx (µ=1.0) results (12.67% in IID, 33.28% light, 59.11% moderate, 4.85% severe) are surprisingly poor and suggest a possible baseline misconfiguration; this should be checked and documented.
[Algorithm 1, V.A.1] It is not stated whether the GMM prototypes are fit on the 80% training split or on the full local dataset. Section V.A.1 describes an 80/20 stratified train/test split, but Algorithm 1 (lines 11-13) says only 'Fit a GMM to local features X_c' without specifying that X_c is restricted to the training split. If the full local data are used, the prototypes encode test-set information and the reported accuracies are inflated. The algorithm and experimental setup must state the split used for prototype fitting.
[IV.D, V.A.6, V.C.6] The number of GMM components is handled inconsistently. Section IV.D says K_c is selected via BIC, while Section V.A.6 and Section V.C.6 assume m_k=3 prototypes per class and use that in the communication-cost calculation. If BIC is used, the number of prototypes varies by class and client, and the advertised '4.37%' upload overhead is not an accurate bound; if m_k=3 is fixed, the text should say so and the BIC statement should be removed. This needs to be clarified for reproducibility.

minor comments (6)

[Table VIII] The table header 'P3E Federated-Learning Hyper-parameters' should read 'FedP3E Federated-Learning Hyper-parameters'.
[Table IX] The Training time column has spacing anomalies (e.g., '8 663.3', '5 788', '5 785.5') and inconsistent decimal places; these should be cleaned up.
[Algorithm 1] Algorithm 1 uses an unspecified evaluation round r*, while the text says the exchange is triggered at round 6 if mean accuracy over rounds 1-5 is below 97%; the algorithm should define r* and state which split is used to compute Accuracy(M_t^G).
[Acknowledgments] The Acknowledgments section contains placeholder text ('This should be a simple paragraph before the References...') that should be removed before submission.
[II] The opening sentence of Section II contains a subject-verb agreement error: 'examine advanced strategies' should be 'examines advanced strategies'.
[Table I] Table I lists IoT-23 with reference [37] while the text in Section II.C refers to [29]; please unify the citation.

Circularity Check

1 steps flagged · score 4.0 of 10

Central accuracy claim is vulnerable to a train/test leakage ambiguity: Algorithm 1 fits GMM prototypes on unqualified 'local features' and adds SMOTE samples derived from them to the training set, while Section V.A.1 defines a held-out test subset.

other [Algorithm 1, lines 11-22; Section V.A.1 (Preprocessing)]
"Fit a GMM to local features Xc ... Extract component means µ(c) j as prototypes ... Apply SMOTE to {ˆµ(c) l } to generate synthetic samples ... Add synthetic data to local dataset Dk ... Each client loads its local CSV partition, scales all features to [0, 1] with Min–Max normalisation, and performs an 80/20 stratified split to create training and test subsets."

Algorithm 1 does not restrict the GMM fit to the training split defined in Section V.A.1. As written, 'local features Xc' is unqualified, so the GMM may be fit on the full local feature matrix including the test subset. The component means become prototypes, are perturbed and aggregated, and SMOTE then synthesizes training samples from those prototypes, which are added to the local dataset before continued training. If the GMM sees the test split, the synthetic training samples encode test-set statistics, so the final test accuracy (95.11-99.71%) partially measures the model's fit to information derived from the test set itself. The reported FedP3E advantage over FedAvg and FedProx would then be forced by construction rather than independently predicted.

full rationale

The only concrete circularity risk in the paper is the unresolved train/test ambiguity in the prototype-generation pipeline: Algorithm 1's unqualified 'local features Xc' and the later 'Add synthetic data to local dataset Dk' can, under the literal reading, inject test-derived information into training, which would make the headline accuracy numbers circular by construction. The privacy claim is unsupported (no sensitivity bound, no DP accounting, no attack evaluation) but that is a correctness gap, not a circularity. The two self-citations ([2], [9]) are contextual and not load-bearing. The communication-cost figure (9.58%) is illustrative arithmetic based on assumed prototype counts, not a circular derivation. If the GMMs are in fact fit only on the training split, the measured performance gains are an empirical property of the method rather than a logical tautology, and the remaining concern is a fairness caveat (baselines receive no synthetic data) rather than circularity. Because the paper leaves the split qualification unstated, the central result is partially vulnerable to a construction-forced outcome, warranting a score of 4 rather than 0.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The framework's measured gains rest on standard tools (GMM, MiniBatch K-Means, SMOTE) plus hand-picked knobs (sigma=0.01, 97% threshold, 10% augmentation) and two unverified assumptions: that the noise provides a privacy guarantee and that prototypes are fit without touching test data. No new physical or algorithmic entities are postulated.

free parameters (5)

Gaussian noise amplitude sigma = 0.01
Tuned empirically 'to balance privacy and utility' (Sec. IV.D); no sensitivity analysis, no differential-privacy bound, and no attack evaluation are derived from it.
Prototype exchange trigger threshold tau = 0.97 (mean accuracy over rounds 1-5)
Set so the one-time exchange fires in the non-IID scenarios and stays dormant in IID (Sec. IV.D, V.A.3); no ablation or sensitivity analysis is reported.
SMOTE augmentation fraction = 10% of training size
Chosen as 'approximately a 10% increase' (Sec. IV.F); no ablation across augmentation sizes is reported.
GMM components per class (mk) = 3 (assumed in the cost estimate)
BIC-selected in principle, but the communication-cost calculation assumes mk=3 per class (Sec. V.C.6); the actual fitted counts are never reported.
Global prototypes per class (mprime_k) = 4 (assumed in the cost estimate)
MiniBatch K-Means cluster count is 'chosen heuristically' (Sec. IV.E); the headline 9.58% overhead figure depends on this assumed value.

assumptions (4)

ad hoc to paper Gaussian noise at sigma=0.01 on [0,1]-scaled prototype means preserves privacy and thwarts sample reconstruction.
Asserted in Sec. IV.D with no threat model, sensitivity bound, differential-privacy accounting, or attack evaluation; the paper's central privacy claim rests on it.
domain assumption Class-conditional GMM means alone carry enough signal for cross-client knowledge transfer.
The pipeline shares only component means and discards covariances, weights, and samples (Sec. IV.D-E); the severe non-IID gains depend on means being sufficient statistics.
domain assumption GMM prototypes are fitted on the 80% training split only, not the held-out 20% test split.
Algorithm 1 fits GMMs to 'local features Xc' without specifying the split, while Sec. V.A.1 defines an 80/20 stratified split. If prototypes use the test split, reported accuracies are inflated by leakage.
domain assumption The reported global accuracy is computed by pooling the three clients' test splits under a consistent rule.
The evaluation protocol for the global model is never stated, yet all numbers in Tables IX-X and Figures 2-9 presuppose some pooling rule; the 49.39% baseline figure is consistent with always predicting the pooled majority class.

how reviews work

0 comments

Cite this review

Pith. "Pith review of FedP3E: Privacy-Preserving Prototype Exchange for Non-IID IoT Malware Detection in Cross-Silo Federated Learning." pith.science (2026). https://pith.science/paper/UGGXGXUL

@misc{pith2026250707258,
  author       = {Pith},
  title        = {Pith review of: FedP3E: Privacy-Preserving Prototype Exchange for Non-IID IoT Malware Detection in Cross-Silo Federated Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UGGXGXUL}},
  note         = {Machine review of arXiv:2507.07258}
}

read the original abstract

As IoT ecosystems continue to expand across critical sectors, they have become prominent targets for increasingly sophisticated and large-scale malware attacks. The evolving threat landscape, combined with the sensitive nature of IoT-generated data, demands detection frameworks that are both privacy-preserving and resilient to data heterogeneity. Federated Learning (FL) offers a promising solution by enabling decentralized model training without exposing raw data. However, standard FL algorithms such as FedAvg and FedProx often fall short in real-world deployments characterized by class imbalance and non-IID data distributions -- particularly in the presence of rare or disjoint malware classes. To address these challenges, we propose FedP3E (Privacy-Preserving Prototype Exchange), a novel FL framework that supports indirect cross-client representation sharing while maintaining data privacy. Each client constructs class-wise prototypes using Gaussian Mixture Models (GMMs), perturbs them with Gaussian noise, and transmits only these compact summaries to the server. The aggregated prototypes are then distributed back to clients and integrated into local training, supported by SMOTE-based augmentation to enhance representation of minority malware classes. Rather than relying solely on parameter averaging, our prototype-driven mechanism enables clients to enrich their local models with complementary structural patterns observed across the federation -- without exchanging raw data or gradients. This targeted strategy reduces the adverse impact of statistical heterogeneity with minimal communication overhead. We evaluate FedP3E on the N-BaIoT dataset under realistic cross-silo scenarios with varying degrees of data imbalance.

Figures

Figures reproduced from arXiv: 2507.07258 by the authors.

**Figure 2.** Comparison of accuracy across 20 rounds under the IID data [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗

**Figure 5.** Evolution of training loss across 20 rounds under the light non-IID [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗

Figures from the paper (4 more)

**Figure 6.** Figure 6: Comparison of accuracy across 20 rounds under the moderate non-IID [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]

**Figure 7.** Figure 7: Evolution of training loss across 20 rounds under the moderate non [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]

**Figure 9.** Figure 9: Evolution of training loss across 20 rounds under the severe non-IID [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]

**Figure 10.** Figure 10: Training time comparison across IID and non-IID scenarios. [PITH_FULL_IMAGE:figures/full_fig_p014_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 38 canonical work pages

[1]

Federated learning for malware detection in IoT devices,

V . Rey, P. M. S. S ´anchez, A. H. Celdr ´an, and G. Bovet, “Federated learning for malware detection in IoT devices,” Computer Networks , vol. 204, p. 108693, 2022

work page 2022
[2]

Deep learning based xiot malware analysis: A comprehensive survey, taxonomy, and research challenges,

R. Darwish, M. Abdelsalam, and S. Khorsandroo, “Deep learning based xiot malware analysis: A comprehensive survey, taxonomy, and research challenges,” Journal of Network and Computer Applications , p. 104258, 2025

work page 2025
[3]

IoT malware surges by 400% in 2023 with US being the most targeted country – Zscaler,

DailyHostNews, “IoT malware surges by 400% in 2023 with US being the most targeted country – Zscaler,” 2023, Accessed: 14 April 2024. [Online]. Available: https://www.linkedin.com/pulse/ iot-malware-surges-400-2023-us-being-most-targeted-country-m9m1f JOURNAL 15

work page 2023
[4]

Machine learning algorithms and frameworks in ransomware detection,

D. Smith, S. Khorsandroo, and K. Roy, “Machine learning algorithms and frameworks in ransomware detection,” IEEE Access , vol. 10, pp. 117 597–117 610, 2022

work page 2022
[5]

An adaptive federated learning scheme with differential privacy preserving,

X. Wu, Y . Zhang, M. Shi, P. Li, R. Li, and N. N. Xiong, “An adaptive federated learning scheme with differential privacy preserving,” Future Generation Computer Systems , vol. 127, pp. 362–372, 2022

work page 2022
[6]

Graph representation feder- ated learning for malware detection in internet of health things,

M. Amjath, S. Henna, and U. Rathnayake, “Graph representation feder- ated learning for malware detection in internet of health things,” Results in Engineering, vol. 25, p. 103651, 2025

work page 2025
[7]

Fs- real: Towards real-world cross-device federated learning,

D. Chen, D. Gao, Y . Xie, X. Pan, Z. Li, Y . Li, B. Ding, and J. Zhou, “Fs- real: Towards real-world cross-device federated learning,” in Proceed- ings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , 2023, pp. 3829–3841

work page 2023
[8]

Sgde: Secure generative data exchange for cross-silo federated learning,

E. Lomurno, A. Archetti, L. Cazzella, S. Samele, L. Di Perna, and M. Matteucci, “Sgde: Secure generative data exchange for cross-silo federated learning,” in Proceedings of the 2022 5th International Confer- ence on Artificial Intelligence and Pattern Recognition , 2022, pp. 205– 214

work page 2022

Show all 44 references

[9]

Comparative analysis of federated learning, deep learning, and traditional machine learning techniques for iot malware detection,

R. Darwish and K. Roy, “Comparative analysis of federated learning, deep learning, and traditional machine learning techniques for iot malware detection,” in 2025 IEEE 4th International Conference on AI in Cybersecurity (ICAIC) . IEEE, 2025, pp. 1–10

2025
[10]

A horizontal federated learning approach to iot malware traffic detection: An empirical evaluation with n-baiot dataset,

P. H. Do, T. D. Le, V . Vishnevsky, A. Berezkin, and R. Kirichek, “A horizontal federated learning approach to iot malware traffic detection: An empirical evaluation with n-baiot dataset,” in 2024 26th International Conference on Advanced Communications Technology (ICACT). IEE...

2024
[11]

A federated learning based botnet detection method for industrial internet of things,

H. Zhou and Z. Sheng, “A federated learning based botnet detection method for industrial internet of things,” in Proceedings of the 8th In- ternational Conference on Cyber Security and Information Engineering , 2023, pp. 282–288

2023
[12]

Distributed optimization for iot attack detection using federated learning and siberian tiger optimizer,

B. B. Gupta, A. Gaurav, W. Alhalabi, V . Arya, E. Alharbi, and K. T. Chui, “Distributed optimization for iot attack detection using federated learning and siberian tiger optimizer,” ICT Express, 2025

2025
[13]

A knowledge transfer- based semi-supervised federated learning for iot malware detection,

X. Pei, X. Deng, S. Tian, L. Zhang, and K. Xue, “A knowledge transfer- based semi-supervised federated learning for iot malware detection,” IEEE Transactions on Dependable and Secure Computing, vol. 20, no. 3, pp. 2127–2143, 2022

2022
[14]

Comprehensive android malware detection based on federated learning architecture,

W. Fang, J. He, W. Li, X. Lan, Y . Chen, T. Li, J. Huang, and L. Zhang, “Comprehensive android malware detection based on federated learning architecture,” IEEE Transactions on Information Forensics and Security, vol. 18, pp. 3977–3990, 2023

2023
[15]

Privacy-preserving malware detection in android-based iot devices through federated markov chains,

G. D’Angelo, E. Farsimadan, M. Ficco, F. Palmieri, and A. Robustelli, “Privacy-preserving malware detection in android-based iot devices through federated markov chains,” Future Generation Computer Sys- tems, vol. 148, pp. 93–105, 2023

2023
[16]

Efficient and lightweight convolutional networks for iot malware detection: A federated learning approach,

M. Abdel-Basset, H. Hawash, K. M. Sallam, I. Elgendi, K. Munasinghe, and A. Jamalipour, “Efficient and lightweight convolutional networks for iot malware detection: A federated learning approach,” IEEE Internet of Things Journal, vol. 10, no. 8, pp. 7164–7173, 2022

2022
[17]

Federated learning with heterogeneous models for on-device malware detection in iot networks,

S. Shukla, S. Rafatirad, H. Homayoun, and S. M. P. Dinakarrao, “Federated learning with heterogeneous models for on-device malware detection in iot networks,” in 2023 Design, Automation & Test in Europe Conference & Exhibition (DATE) . IEEE, 2023, pp. 1–6

2023
[18]

Sim-fed: Secure iot malware detection model with federated learning,

M. Nobakht, R. Javidan, and A. Pourebrahimi, “Sim-fed: Secure iot malware detection model with federated learning,” Computers and Electrical Engineering, vol. 116, p. 109139, 2024

2024
[19]

Federated learning-based ran- somware detection via indicators of compromise,

S. Koike, H. Tanaka, and M. Maeda, “Federated learning-based ran- somware detection via indicators of compromise,” 2024

2024
[20]

Privacy-preserving federated learning approach for distributed malware attacks with in- termittent clients and image representation,

F. Ullah, G. Srivastava, S. Ullah, and L. Mostarda, “Privacy-preserving federated learning approach for distributed malware attacks with in- termittent clients and image representation,” IEEE Transactions on Consumer Electronics, vol. 70, no. 1, pp. 4585–4596, 2023

2023
[21]

Client selection for federated learning with non-iid data in mobile edge computing,

W. Zhang, X. Wang, P. Zhou, W. Wu, and X. Zhang, “Client selection for federated learning with non-iid data in mobile edge computing,” IEEE Access, vol. 9, pp. 24 462–24 474, 2021

2021
[22]

Resource-efficient federated learning with non-iid data: An auction theoretic approach,

E. Seo, D. Niyato, and E. Elmroth, “Resource-efficient federated learning with non-iid data: An auction theoretic approach,” IEEE Internet of Things Journal, vol. 9, no. 24, pp. 25 506–25 524, 2022

2022
[23]

Fedsld: Federated learning with shared label distribu- tion for medical image classification,

J. Luo and S. Wu, “Fedsld: Federated learning with shared label distribu- tion for medical image classification,” in 2022 IEEE 19th International Symposium on Biomedical Imaging (ISBI) . IEEE, 2022, pp. 1–5

2022
[24]

Fednse: Optimal node selection for federated learning with non-iid data,

S. Bansal, M. Bansal, R. Verma, R. Shorey, and H. Saran, “Fednse: Optimal node selection for federated learning with non-iid data,” in 2023 15th International Conference on COMmunication Systems & NETworkS (COMSNETS). IEEE, 2023, pp. 713–721

2023
[25]

K-fl: Kalman filter- based clustering federated learning method,

H. Kim, B. Kim, Y . Kim, C. You, and H. Park, “K-fl: Kalman filter- based clustering federated learning method,” IEEE Access, vol. 11, pp. 36 097–36 105, 2023

2023
[26]

Clustered federated multitask learning on non-iid data with enhanced privacy,

J. Shu, T. Yang, X. Liao, F. Chen, Y . Xiao, K. Yang, and X. Jia, “Clustered federated multitask learning on non-iid data with enhanced privacy,”IEEE Internet of Things Journal, vol. 10, no. 4, pp. 3453–3467, 2022

2022
[27]

{IoTPOT}: analysing the rise of {IoT} compromises,

Y . M. P. Pa, S. Suzuki, K. Yoshioka, T. Matsumoto, T. Kasama, and C. Rossow, “ {IoTPOT}: analysing the rise of {IoT} compromises,” in 9th USENIX Workshop on Offensive Technologies (WOOT 15) , 2015

2015
[28]

Bot-iot dataset - unsw research,

U. of New South Wales (UNSW), “Bot-iot dataset - unsw research,” https://research.unsw.edu.au/projects/bot-iot-dataset, accessed: 03 Jan 2025

2025
[29]

Iot-23 dataset: A labeled dataset for iot malware and benign traffic,

S. Laboratory, “Iot-23 dataset: A labeled dataset for iot malware and benign traffic,” https://www.stratosphereips.org/datasets-iot23, accessed: 03 Jan 2025

2025
[30]

Drebin: Effective and explainable detection of android malware in your pocket

D. Arp, M. Spreitzenbarth, M. Hubner, H. Gascon, K. Rieck, and C. Siemens, “Drebin: Effective and explainable detection of android malware in your pocket.” in Ndss, vol. 14, 2014, pp. 23–26

2014
[31]

Microsoft malware classification challenge,

R. Ronen, M. Radu, C. Feuerstein, E. Yom-Tov, and M. Ah- madi, “Microsoft malware classification challenge,” arXiv preprint arXiv:1802.10135, 2018

2018 arXiv
[32]

Malware images: visualization and automatic classification,

L. Nataraj, S. Karthikeyan, G. Jacob, and B. S. Manjunath, “Malware images: visualization and automatic classification,” in Proceedings of the 8th international symposium on visualization for cyber security , 2011, pp. 1–7

2011
[33]

Dissecting android malware: Characterization and evolution,

Y . Zhou and X. Jiang, “Dissecting android malware: Characterization and evolution,” in2012 IEEE symposium on security and privacy. IEEE, 2012, pp. 95–109

2012
[34]

A detailed analysis of the kdd cup 99 data set,

M. Tavallaee, E. Bagheri, W. Lu, and A. A. Ghorbani, “A detailed analysis of the kdd cup 99 data set,” in 2009 IEEE symposium on computational intelligence for security and defense applications . Ieee, 2009, pp. 1–6

2009
[35]

Intrusion detection system for healthcare systems using medical and network data: A comparison study,

A. A. Hady, A. Ghubaish, T. Salman, D. Unal, and R. Jain, “Intrusion detection system for healthcare systems using medical and network data: A comparison study,” IEEE Access, vol. 8, pp. 106 576–106 584, 2020

2020
[36]

N-baiot—network-based detection of iot botnet attacks using deep autoencoders,

Y . Meidan, M. Bohadana, Y . Mathov, Y . Mirsky, A. Shabtai, D. Breiten- bacher, and Y . Elovici, “N-baiot—network-based detection of iot botnet attacks using deep autoencoders,” IEEE Pervasive Computing , vol. 17, no. 3, pp. 12–22, 2018

2018
[37]

Iot-23: A labeled dataset with malicious and benign iot network traffic,

S. Garcia, A. Parmisano, and M. J. Erquiaga, “Iot-23: A labeled dataset with malicious and benign iot network traffic,” (No Title), 2020

2020
[38]

Towards the development of realistic botnet dataset in the internet of things for network forensic analytics: Bot-iot dataset,

N. Koroniotis, N. Moustafa, E. Sitnikova, and B. Turnbull, “Towards the development of realistic botnet dataset in the internet of things for network forensic analytics: Bot-iot dataset,” Future Generation Computer Systems, vol. 100, pp. 779–796, 2019

2019
[39]

Cross-silo federated learning: Chal- lenges and opportunities,

C. Huang, J. Huang, and X. Liu, “Cross-silo federated learning: Chal- lenges and opportunities,” arXiv preprint arXiv:2206.12949 , 2022

2022 arXiv
[40]

Federated learning with non-iid data: A survey,

Z. Lu, H. Pan, Y . Dai, X. Si, and Y . Zhang, “Federated learning with non-iid data: A survey,” IEEE Internet of Things Journal , 2024

2024
[41]

Client specific dynamic aggregation for non-iid federated learning,

V . Altomare, D. Thakur, A. Guzzo, and F. Piccialli, “Client specific dynamic aggregation for non-iid federated learning,” in 2024 IEEE International Conference on Big Data (BigData) . IEEE, 2024, pp. 7622–7631

2024
[42]

Communication-efficient learning of deep networks from decentralized data,

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

2017
[43]

Federated optimization in heterogeneous networks,

T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, and V . Smith, “Federated optimization in heterogeneous networks,” Proceedings of Machine learning and systems , vol. 2, pp. 429–450, 2020

2020
[44]

A non-parametric view of fedavg and fedprox: Beyond stationary points,

L. Su, J. Xu, and P. Yang, “A non-parametric view of fedavg and fedprox: Beyond stationary points,” arXiv preprint arXiv:2106.15216 , 2021

2021 arXiv

Pith tools

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

[1] [1]

Federated learning for malware detection in IoT devices,

V . Rey, P. M. S. S ´anchez, A. H. Celdr ´an, and G. Bovet, “Federated learning for malware detection in IoT devices,” Computer Networks , vol. 204, p. 108693, 2022

work page 2022

[2] [2]

Deep learning based xiot malware analysis: A comprehensive survey, taxonomy, and research challenges,

R. Darwish, M. Abdelsalam, and S. Khorsandroo, “Deep learning based xiot malware analysis: A comprehensive survey, taxonomy, and research challenges,” Journal of Network and Computer Applications , p. 104258, 2025

work page 2025

[3] [3]

IoT malware surges by 400% in 2023 with US being the most targeted country – Zscaler,

DailyHostNews, “IoT malware surges by 400% in 2023 with US being the most targeted country – Zscaler,” 2023, Accessed: 14 April 2024. [Online]. Available: https://www.linkedin.com/pulse/ iot-malware-surges-400-2023-us-being-most-targeted-country-m9m1f JOURNAL 15

work page 2023

[4] [4]

Machine learning algorithms and frameworks in ransomware detection,

D. Smith, S. Khorsandroo, and K. Roy, “Machine learning algorithms and frameworks in ransomware detection,” IEEE Access , vol. 10, pp. 117 597–117 610, 2022

work page 2022

[5] [5]

An adaptive federated learning scheme with differential privacy preserving,

X. Wu, Y . Zhang, M. Shi, P. Li, R. Li, and N. N. Xiong, “An adaptive federated learning scheme with differential privacy preserving,” Future Generation Computer Systems , vol. 127, pp. 362–372, 2022

work page 2022

[6] [6]

Graph representation feder- ated learning for malware detection in internet of health things,

M. Amjath, S. Henna, and U. Rathnayake, “Graph representation feder- ated learning for malware detection in internet of health things,” Results in Engineering, vol. 25, p. 103651, 2025

work page 2025

[7] [7]

Fs- real: Towards real-world cross-device federated learning,

D. Chen, D. Gao, Y . Xie, X. Pan, Z. Li, Y . Li, B. Ding, and J. Zhou, “Fs- real: Towards real-world cross-device federated learning,” in Proceed- ings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , 2023, pp. 3829–3841

work page 2023

[8] [8]

Sgde: Secure generative data exchange for cross-silo federated learning,

E. Lomurno, A. Archetti, L. Cazzella, S. Samele, L. Di Perna, and M. Matteucci, “Sgde: Secure generative data exchange for cross-silo federated learning,” in Proceedings of the 2022 5th International Confer- ence on Artificial Intelligence and Pattern Recognition , 2022, pp. 205– 214

work page 2022