Adaptive Pruning of Deep Neural Networks for Resource-Aware Embedded Intrusion Detection on the Edge

Alexandre Broggi; Gokhan Kul; Lance Fiondella; Nathaniel Bastian

REVIEW 4 major objections 6 minor 20 references

Adaptive Pruning of Deep Neural Networks for Resource-Aware Embedded Intrusion Detection on the Edge

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper finds that most CNN-oriented pruning algorithms lose significant accuracy when applied to a fully connected network on the ACI IoT intrusion detection dataset, and that ThiNet is the best practical pruner for this setting.

desk verdict Honest empirical benchmark with a load-bearing DAIS implementation gap; the ThiNet result likely stands, but the 'many methods don't transfer' claim needs caveats. read the letter →

arxiv 2505.14592 v1 pith:EFGB3KQB submitted 2025-05-20 cs.LG cs.CR

Alexandre Broggi , Nathaniel Bastian , Lance Fiondella , Gokhan Kul This is my paper · ORCID

classification cs.LGcs.CR

keywords networkintrusiondetectiondeepneuralpruningIoTsystemsonchipfullyconnectedmodelcompressionThiNetedgecomputing

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

Pruning reduces a trained network's size so it can run on embedded hardware, but pruning methods are usually designed for convolutional networks. This paper asks whether those methods still work when transplanted to a much simpler fully connected network trained on the ACI-IOT-2023-payload intrusion-detection dataset, testing eight approaches across pruning levels from 99% down to 4% of the original filters. The paper's central finding is that most CNN-era algorithms do not transfer: accuracy drops sharply past a critical pruning point, runtimes become impractical, or results are unstable across random trials. ThiNet, a heuristic that removes filters with the lowest contribution to the next layer, emerges as the best practical method, while the authors' own Iterative-Theseus scores highest on F1 but at a far greater time cost. The authors conclude that the field cannot assume CNN pruning methods will work on other network types, especially for resource-constrained edge intrusion detection.

What carries the argument

The argument rests on an empirical comparison protocol: a fixed fully connected network (two outer linear layers plus 27 hidden layers with leaky ReLU and dropout) trained on the ACI-IOT-2023-payload dataset, then pruned by each method at ten target percentages from 99% down to 4%. Results are scaled relative to the original F1 score and parameter count, with each configuration run three times and averaged. The key contrast is between 'Hindered' algorithms, whose F1 falls past a critical pruning point, and 'Helped' algorithms, which improve as more is removed. Within this setup, ThiNet's mechanism—ranking each filter by the total activation it contributes to the next layer and pruning the lowest-ranked filters—is what best preserves accuracy per unit of pruning time.

What would settle it

Re-run DAIS on this fully connected ACI model with all three regularizers enabled and the original documented training cycle; if its F1 then matches or beats ThiNet across pruning percentages, the paper's conclusion that DAIS does not transfer to fully connected networks would be refuted.

Watch

Extended reading notes

Core claim

The paper claims that pruning algorithms designed for convolutional neural networks largely fail to generalize to a fully connected network on the ACI IoT intrusion-detection benchmark. Of the eight methods tested, ThiNet offers the best trade-off between F1 score and pruning time, with a retrained-from-scratch model close behind. Two algorithms are 'helped' by aggressive pruning—DAIS and BERT-Theseus—but DAIS is highly variable between runs and BERT-Theseus only becomes competitive when large layer blocks are removed. The authors conclude that most CNN pruning methods cannot be transferred to fully connected networks without significant loss of potency, and interpret the strong performance of retraining as evidence that the original model was oversized for the dataset.

Load-bearing premise

The evaluation assumes that the re-implementations of ADMM and DAIS faithfully capture the original algorithms, even though DAIS was run with only one of its three regularizers and an unrecoverable training schedule, and ADMM's mask was applied after the linear layer rather than during matrix multiplication.

Editorial extensions

If this is right

On a fully connected network trained on ACI IoT data, the practical choice is ThiNet: it keeps high F1 scores at moderate pruning levels and finishes pruning in a fraction of Iterative-Theseus's time.
Simply retraining a model resized to the pruned dimension is nearly as good as most dedicated pruning methods, suggesting the original 29-layer model was oversized for this dataset.
BERT-Theseus is only useful when removing large continuous chunks of layers; at light pruning it underperforms because equal-sized replacement modules must be trained from scratch.
Most CNN-oriented pruning algorithms cannot be moved to fully connected networks without significant loss of accuracy, which matters for deploying pruned detectors on edge hardware.
DAIS in its tested form is unreliable on non-residual, fully connected networks, a consequence of the missing regularizers and an unrecoverable training schedule.

Reading between the lines

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

If the transfer failure is real, it suggests that the implicit redundancy structure in convolutional networks (spatial correlation across channels) is what many pruning regularizers rely on; fully connected layers, where each filter has no spatial structure, remove that crutch.
A decisive follow-up would be to run ThiNet and a full-fidelity DAIS/ADMM on the same fully connected model to separate algorithmic failure from implementation drift.
For edge deployment, the paper's results imply that a simple combination of random-structured pruning plus retraining (or no pruning at all with a smaller architecture) may be a cheaper baseline than any learned pruning method for this dataset.

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: this is a solid, honest empirical comparison that is worth engaging, but the broad conclusion that CNN-oriented pruning methods fail on a fully connected NIDS model is only as strong as the most questionable implementation in the batch. The DAIS port uses one of three regularizers and a training cycle that could not be reproduced because the reference sheet was missing. That alone makes the 'many do not generalize' claim shaky. Still, the ThiNet recommendation and the practical observations are likely to survive.

What the paper does well: it runs eight methods (including two baselines) on a real IoT intrusion detection dataset, on two model sizes, across ten pruning levels, with three trials per cell and standard deviations reported. Code and data are on Github. The authors are unusually transparent about their deviations: they disclose the ADMM mask placement, the DAIS regularizer situation, and the BERT-Theseus percentage approximation. That transparency is a real credit. The finding that Iterative-Theseus performs well but is far too slow, and that a from-scratch retrain is competitive, is actionable for practitioners. The observation that their initial model was oversized is a sensible post-hoc interpretation.

Soft spots, in order: the DAIS implementation fidelity is the load-bearing one. With DAIS excluded or labeled as an approximation, the 'many methods' claim reduces to ADMM (whose mask-before-or-after-linear-layer change is at least mathematically plausible to be neutral) plus whatever BERT-Theseus is doing. The statistics are also thin: three runs with a z-score-based confidence interval assumes normality, which the authors themselves flag. That is a minor issue for a benchmark, not a fatal one. The BERT-Theseus pruning percentages are approximate, and the reported parameter counts sometimes increase at the end, which muddies the headline numbers a bit.

Recommendation: send it to peer review. The authors have a reproducible artifact and an honest write-up. The right referee will push them to either re-run DAIS faithfully or soften the generalization claim, and to add a couple more seeds or nonparametric confidence bounds. This is exactly the kind of paper that improves with review rather than being a desk-reject.

Referee Report

4 major / 6 minor

Summary. The paper compares eight pruning approaches—ADMM-joint, BERT-Theseus, DAIS, Iterative Theseus, ThiNet, random structured pruning, and complete retraining/recreation—applied to a fully connected network trained on the ACI IoT intrusion detection dataset. It reports F1 scores, parameter counts, and pruning times for two model sizes across a grid of requested pruning percentages, proposes Iterative Theseus as a layer-replacement baseline, and concludes that ThiNet is the best practical method while most CNN-oriented pruning methods do not transfer to this fully connected setting.

Significance. If the central finding were fully supported, the paper would offer useful practical guidance for deploying pruned intrusion detection models on resource-constrained edge hardware. The authors provide code and result data, run two model sizes, and include control baselines (random pruning and retraining from scratch), which are appropriate experimental ingredients. However, the broad transferability claim is substantially weaker than the abstract suggests because the DAIS implementation is disclosed to be incomplete, the BERT-Theseus pruning percentages are approximate, and the statistical analysis rests on three runs with normal-theory error bars. The specific recommendation that ThiNet is competitive is more credible than the negative generalization claim, but even that recommendation needs to be expressed relative to a clearer criterion.

major comments (4)

[Section III-B and Table III] The central negative claim that many CNN-oriented pruning methods do not generalize is materially supported by the DAIS results, which reach F1 values as low as 0.013 at 74% retained parameters on the large model. However, the manuscript itself discloses that the DAIS implementation used only one of its three regularizers (the residual-block regularizer was inapplicable and the lasso regularizer was disabled) and that the DARTS-style alternating training cycle could not be reproduced because the reference sheet "was unable to be located." As described, this is not a faithful port of DAIS from [16], so the DAIS failure cannot be cited as evidence that the original algorithm fails to transfer. Please either implement the full method with all applicable regularizers and the documented training schedule, or revise the conclusion to explicitly restrict the claim to the implemented variants and remove DAIS from the list of algorithms whose transferability is being assessed.
[Section V and Table III (BERT-Theseus rows)] The mapping from requested pruning percentage to actual model size is approximate for BERT-Theseus, and the table shows identical parameter counts and identical F1 values across four consecutive requested percentages (small model: 329k parameters and F1 0.106 for percentages 0.99 through 0.62; large model: 1233k parameters and F1 0.041 for the same range). These rows are not independent measurements of distinct pruning degrees, yet Figures 3 and 4 connect them as a single curve. Please report the effective per-layer sizes actually produced by the block-selection heuristic and restrict the "Helped" characterization to configurations that genuinely differ in model size.
[Section V, error-bar formula and Table III] The error bars are computed as 1.960*sigma/sqrt(3) under a normal assumption, which the authors themselves acknowledge may be incorrect for n=3. The standard deviations are large precisely in rows that drive the conclusions (e.g., DAIS small model at 0.08: F1 0.343 +/- 0.52; random structured small model at 0.25: F1 0.345 +/- 0.41; recreation small model at 0.08: F1 0.574 +/- 0.41). The text also makes ordinal claims, such as the order in which the "Hindered" algorithms start to degrade, without any significance test or paired comparison. Please provide per-run data or appropriate small-sample intervals and state which pairwise differences are actually supported by the data.
[Section VI and Table III] The concluding recommendation that ThiNet is "the best algorithm" for this dataset is not uniformly supported by the reported F1 values. On the large model, Iterative Theseus achieves higher F1 than ThiNet at percentages 0.99 (0.800 vs 0.789), 0.49 (0.745 vs 0.660), and 0.25 (0.702 vs 0.523), while Recreation is close at several points. Since the relative ranking depends on pruning degree, time budget, and whether F1 or efficiency is prioritized, the recommendation should be stated as a Pareto-style or criterion-dependent conclusion rather than a single global winner.

minor comments (6)

[Section III-C, footnote 1] The footnote correctly observes that Gou et al. [18] does not actually classify BERT-Theseus, but the main text still attributes the classification to that survey; please reconcile the text with the footnote.
[Section IV] The architecture description is ambiguous: "a pair of linear layers surrounding 27 fully connected linear layers" does not make the total layer count precise, and the statement that filters per hidden layer increase by 1 per layer should be accompanied by an explicit schedule for both the 175-filter and 75-filter models.
[Section V] The phrase "at just around 0.2%" appears to mean a pruning fraction of 0.2 (or 20% of filters retained), not 0.2%; please correct the units to avoid confusion with the x-axis of the figures.
[Table II] The use of single and double asterisks in the grouped and under-sampled columns is explained only implicitly; please add a legend or table note defining each symbol and the grouping scheme.
[References] Reference [9] is a non-archival blog post used for a general claim about weight-based pruning; consider replacing it with a peer-reviewed source or clearly labeling it as an online resource.
[Throughout] There are several typographical inconsistencies, including "regulairizers" (Section III-A), "thinet" vs "ThiNet", "BERT-Thesesus" (Section VI), and "loose" for "lose"; a careful proofreading pass is needed.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: this is an empirical comparison of externally defined pruning methods against a public dataset; the disclosed caveats are implementation-fidelity risks, not circularity.

full rationale

The paper's central claim — that CNN-oriented pruning methods such as ADMM, DAIS, and ThiNet transfer imperfectly to a fully connected network on the ACI IoT dataset, and that ThiNet is the practical best — is an empirical benchmark result. The pruning rules in Section III are taken from external sources [12], [16], [17] and are not defined in terms of the paper's own outcome measure; the F1 scores in Section V and Table III are measured outcomes, not constructed outputs. No parameter is fitted to a subset of the data and then reported as a prediction of a closely related quantity, and no equation defines a claimed derivation in terms of its own output. The only connection to the authors' prior work is the ACI dataset itself (reference [3], co-authored by N. Bastian), which is a public benchmark independent of the pruning comparison, and reference [8], a hardware co-design paper not used as evidence for the transferability conclusion. The paper's own disclosures in Section III-B — that DAIS was run with only one of its three regularizers and that the DARTS-style training-cycle reference sheet was unable to be located — are genuine threats to implementation fidelity and therefore to the strength of the 'many methods do not generalize' conclusion, but they are external-validity and correctness concerns, not circular reasoning. The footnote correcting the BERT-Theseus classification in [18] is a citation error, not a circular step. Accordingly, no load-bearing step reduces to its own inputs, and the correct circularity finding is no significant circularity (0).

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the choice of the ACI IoT dataset, the architecture (27 fully connected layers with 175/75 filters), the specific implementations of each pruning method, and the use of macro-averaged F1 with three runs. The most fragile link is the fidelity of the modified implementations, especially DAIS.

free parameters (6)

learning_rate = 0.0009
Chosen for training the initial model; affects all pruning results but is not fitted to the pruning comparison.
dropout = 0.0002
Dropout rate used during training; an ad hoc choice.
hidden_layer_filter_counts = 175 and 75 per layer, increasing by 1 per layer
Model width chosen by a short random search and then reduced; affects the redundancy available for pruning.
number_of_hidden_layers = 27
Chosen by a short random search; the paper notes thinet time scaled poorly with layers, so it was reduced.
pruning_percentage_grid = 0.99 down to 0.04
The set of pruning degrees tested; chosen by the authors, and the BERT-Theseus mapping to these percentages is approximate.
z_score_for_error_bars = 1.960
Used to compute 95% confidence intervals assuming a normal distribution with n=3; the authors acknowledge this assumption may be incorrect.

assumptions (4)

domain assumption Macro-averaged F1 is an appropriate metric for the imbalanced intrusion detection task.
The dataset is highly imbalanced (e.g., ICMP Flood 58 vs Benign 601,868); the paper averages class F1 but does not justify this choice over precision-recall AUC or cost-sensitive metrics.
domain assumption The ACI-IOT-2023-payload dataset is a valid and representative benchmark for IoT network intrusion detection.
The dataset is created by a co-author (Bastian) and used without external validation; the paper relies on it as ground truth.
ad hoc to paper The modified implementations of DAIS, ADMM, and BERT-Theseus preserve the behavior of the original algorithms.
DAIS uses only one of three regularizers and an unspecified training cycle; ADMM's mask is applied after the linear layer rather than during matrix multiplication; BERT-Theseus replaces layers in approximate batches. The paper does not demonstrate that these changes are neutral.
domain assumption Three repeated runs per condition are sufficient to estimate performance and rank algorithms.
The paper computes error bars with n=3 and a normal assumption, which it admits may be incorrect; many entries show zero variance due to degenerate model collapse.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Adaptive Pruning of Deep Neural Networks for Resource-Aware Embedded Intrusion Detection on the Edge." pith.science (2026). https://pith.science/paper/EFGB3KQB

@misc{pith2026250514592,
  author       = {Pith},
  title        = {Pith review of: Adaptive Pruning of Deep Neural Networks for Resource-Aware Embedded Intrusion Detection on the Edge},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EFGB3KQB}},
  note         = {Machine review of arXiv:2505.14592}
}

read the original abstract

Artificial neural network pruning is a method in which artificial neural network sizes can be reduced while attempting to preserve the predicting capabilities of the network. This is done to make the model smaller or faster during inference time. In this work we analyze the ability of a selection of artificial neural network pruning methods to generalize to a new cybersecurity dataset utilizing a simpler network type than was designed for. We analyze each method using a variety of pruning degrees to best understand how each algorithm responds to the new environment. This has allowed us to determine the most well fit pruning method of those we searched for the task. Unexpectedly, we have found that many of them do not generalize to the problem well, leaving only a few algorithms working to an acceptable degree.

Figures

Figures reproduced from arXiv: 2505.14592 by the authors.

**Figure 2.** The model structure used for training Count Grouped Full Under- Grouped UnderClassification Count sampled Count sampled Benign 601, 868 5, 800 601, 868 11, 830 DNS Flood 18, 577 5, 800 20, 816* 11, 830* Dictionary Attack 4, 645 4, 645 4, 645 4, 645 Slowloris 2, 974 2, 974 2, 974 2, 974 SYN Flood 2, 113 2, 113 * * Port Scan 582 582 1, 183** 1, 183** Vulnerability Scan 445 445 ** ** OS Scan 156 156 ** ** UDP Flood 68… view at source ↗

**Figure 3.** Big pruning model scaled such that the original run is at (1, 1) V. RESULTS We ran each algorithm three times on each of ten pruning percentages. We are defining ’pruning percentages’ to be the estimated percentage of filters for each hidden layer after pruning has been applied, or an estimated replacement. For example, ADDM joint pruning has both ’k’ and ’percent’ options, so each is scaled by the pruning percent. … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

20 extracted references · 13 canonical work pages

[16]

Dais: Automatic channel pruning via differentiable annealing indicator search,

Y . Guan, N. Liu, P. Zhao, Z. Che, K. Bian, Y . Wang, and J. Tang, “Dais: Automatic channel pruning via differentiable annealing indicator search,” IEEE Transactions on Neural Networks and Learning Systems , vol. 34, no. 12, pp. 9847–9858, 2022

work page 2022
[1]

Natural language processing,

S. C. Fanni, M. Febi, G. Aghakhanyan, and E. Neri, “Natural language processing,” in Introduction to Artificial Intelligence . Springer, 2023, pp. 87–99

work page 2023
[2]

Task-oriented feature dis- tillation,

L. Zhang, Y . Shi, Z. Shi, K. Ma, and C. Bao, “Task-oriented feature dis- tillation,” Advances in Neural Information Processing Systems , vol. 33, pp. 14 759–14 771, 2020

work page 2020
[3]

Aci iot network traffic dataset 2023,

N. Bastian, D. Bierbrauer, M. McKenzie, and E. Nack, “Aci iot network traffic dataset 2023,” 2023. [Online]. Available: https: //dx.doi.org/10.21227/qacj-3x32

doi:10.21227/qacj-3x32 2023
[4]

Kitsune: An ensemble of autoencoders for online network intrusion detection,

Y . Mirsky, T. Doitshman, Y . Elovici, and A. Shabtai, “Kitsune: An ensemble of autoencoders for online network intrusion detection,”

work page
[5]

Bert-of-theseus: Compressing bert by progressive module replacing,

C. Xu, W. Zhou, T. Ge, F. Wei, and M. Zhou, “Bert-of-theseus: Compressing bert by progressive module replacing,” 2020. [Online]. Available: https://arxiv.org/abs/2002.02925

arXiv 2020
[6]

Structured pruning for deep convolutional neural networks: A survey,

Y . He and L. Xiao, “Structured pruning for deep convolutional neural networks: A survey,”IEEE transactions on pattern analysis and machine intelligence, vol. 46, pp. 2900–2919, 2023

work page 2023
[7]

Applications of machine learning and deep learning in securing internet of battlefield things: A futuristic perspective,

S. Joshi, A. Thakar, and C. Patel, “Applications of machine learning and deep learning in securing internet of battlefield things: A futuristic perspective,” in 2023 10th International Conference on Computing for Sustainable Global Development (INDIACom) . IEEE, 2023, pp. 333– 338

work page 2023

Show all 20 references

[8]

Co-design of decision trees for network intrusion detection at the edge on digital vs. analog hardware,

J. Riem, L. Zhang, J. Chen, H. Mackay, T. Lan, N. D. Bastian, and G. C. Adam, “Co-design of decision trees for network intrusion detection at the edge on digital vs. analog hardware,” in MILCOM 2024 - 2024 IEEE Military Communications Conference (MILCOM) , 2024, pp. 39–44

2024
[9]

How to forget jenny’s phone num- ber,

V . Nielson, “How to forget jenny’s phone num- ber,” 2023. [Online]. Available: https://vianielson.substack.com/p/ af66fa92-308b-4141-8ccd-f35cc01539a3

2023
[10]

An introduction to computing with neural nets,

R. P. Lippmann, “An introduction to computing with neural nets,” ACM SIGARCH Computer Architecture News, vol. 16, no. 1, pp. 7–25, 1988

1988
[11]

Learning both weights and connections for efficient neural network,

S. Han, J. Pool, J. Tran, and W. Dally, “Learning both weights and connections for efficient neural network,” in Advances in Neural Information Processing Systems , C. Cortes, N. Lawrence, D. Lee, M. Sugiyama, and R. Garnett, Eds., vol. 28. Curran Associates, Inc., 2015, pp. 1...

2015
[12]

Dnn compression by admm-based joint pruning,

G. Lee and K. Lee, “Dnn compression by admm-based joint pruning,” Knowledge-Based Systems , vol. 239, p. 107988, 2022. [Online]. Available: https://www.sciencedirect.com/science/article/pii/ S0950705121011047

2022
[13]

Distilling the knowledge in a neural network,

G. Hinton, O. Vinyals, and J. Dean, “Distilling the knowledge in a neural network,” 2015. [Online]. Available: https://arxiv.org/abs/1503.02531

2015 arXiv
[14]

Compressing deep convolutional networks using vector quantization,

Y . Gong, L. Liu, M. Yang, and L. Bourdev, “Compressing deep convolutional networks using vector quantization,” 2014. [Online]. Available: https://arxiv.org/abs/1412.6115

2014 arXiv
[15]

L1unstructured,

T. P. Foundation, “L1unstructured,” The PyTorch Foundation, october 2024, documentation for pytorch v2.5. [Online]. Available: https://pytorch.org/docs/stable/generated/torch.nn.utils.prune. L1Unstructured.html#torch.nn.utils.prune.L1Unstructured

2024
[17]

Thinet: A filter level pruning method for deep neural network compression,

J.-H. Luo, J. Wu, and W. Lin, “Thinet: A filter level pruning method for deep neural network compression,” in Proceedings of the IEEE international conference on computer vision . IEEE, 2017, pp. 5058– 5066

2017
[18]

Knowledge distillation: A survey,

J. Gou, B. Yu, S. J. Maybank, and D. Tao, “Knowledge distillation: A survey,” International Journal of Computer Vision , vol. 129, no. 6, pp. 1789–1819, 2021

2021
[19]

Darts: Differentiable architecture search,

H. Liu, K. Simonyan, and Y . Yang, “Darts: Differentiable architecture search,” 2018. [Online]. Available: https://arxiv.org/abs/1806.09055 APPENDIX small big Parameters F1 Score Standard Time Parameters F1 Score Standard Time PruningSelection Percent Deviation (Log) Deviation...

2018 arXiv
[2018]

Available: https://arxiv.org/abs/1802.09089

[Online]. Available: https://arxiv.org/abs/1802.09089

arXiv

Pith tools

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

[1] [16]

Dais: Automatic channel pruning via differentiable annealing indicator search,

Y . Guan, N. Liu, P. Zhao, Z. Che, K. Bian, Y . Wang, and J. Tang, “Dais: Automatic channel pruning via differentiable annealing indicator search,” IEEE Transactions on Neural Networks and Learning Systems , vol. 34, no. 12, pp. 9847–9858, 2022

work page 2022

[2] [1]

Natural language processing,

S. C. Fanni, M. Febi, G. Aghakhanyan, and E. Neri, “Natural language processing,” in Introduction to Artificial Intelligence . Springer, 2023, pp. 87–99

work page 2023

[3] [2]

Task-oriented feature dis- tillation,

L. Zhang, Y . Shi, Z. Shi, K. Ma, and C. Bao, “Task-oriented feature dis- tillation,” Advances in Neural Information Processing Systems , vol. 33, pp. 14 759–14 771, 2020

work page 2020

[4] [3]

Aci iot network traffic dataset 2023,

N. Bastian, D. Bierbrauer, M. McKenzie, and E. Nack, “Aci iot network traffic dataset 2023,” 2023. [Online]. Available: https: //dx.doi.org/10.21227/qacj-3x32

doi:10.21227/qacj-3x32 2023

[5] [4]

Kitsune: An ensemble of autoencoders for online network intrusion detection,

Y . Mirsky, T. Doitshman, Y . Elovici, and A. Shabtai, “Kitsune: An ensemble of autoencoders for online network intrusion detection,”

work page

[6] [5]

Bert-of-theseus: Compressing bert by progressive module replacing,

C. Xu, W. Zhou, T. Ge, F. Wei, and M. Zhou, “Bert-of-theseus: Compressing bert by progressive module replacing,” 2020. [Online]. Available: https://arxiv.org/abs/2002.02925

arXiv 2020

[7] [6]

Structured pruning for deep convolutional neural networks: A survey,

Y . He and L. Xiao, “Structured pruning for deep convolutional neural networks: A survey,”IEEE transactions on pattern analysis and machine intelligence, vol. 46, pp. 2900–2919, 2023

work page 2023

[8] [7]

Applications of machine learning and deep learning in securing internet of battlefield things: A futuristic perspective,

S. Joshi, A. Thakar, and C. Patel, “Applications of machine learning and deep learning in securing internet of battlefield things: A futuristic perspective,” in 2023 10th International Conference on Computing for Sustainable Global Development (INDIACom) . IEEE, 2023, pp. 333– 338

work page 2023