arxiv: 2605.13922 · v1 · submitted 2026-05-13 · 💻 cs.CR · cs.LG· stat.CO

Recognition: no theorem link

XAI and Statistical Analysis for Reliable Intrusion Detection in the UAVIDS-2025 Dataset: From Tree to Hybrid and Tabular DNN Ensembles

Iakovos-Christos Zarkadis , Christos Douligeris

Authors on Pith no claims yet

Pith reviewed 2026-05-15 05:45 UTC · model grok-4.3

classification 💻 cs.CR cs.LGstat.CO

keywords UAV intrusion detectionexplainable AISHAPXGBoostWormhole attackBlackhole attackdensity support intersectionstatistical analysis

0 comments

The pith

XGBoost with SHAP and statistical tests shows density support intersections cause misclassifications in Wormhole and Blackhole attacks on UAVIDS-2025

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper evaluates tree ensembles, deep neural networks, hybrid stacking models, and tabular DNN ensembles for intrusion detection in UAV networks on the UAVIDS-2025 dataset under stratified 10-fold cross validation. XGBoost is identified as the strongest model, after which SHAP values are used to map global and local feature importances and expose how each attack modifies traffic to resemble normal patterns. Kernel density estimates are then compared visually and quantitatively with Westfall-Young permutation tests and Jensen-Shannon distances to establish that density support intersections between normal and attack classes drive the observed false predictions specifically for Wormhole and Blackhole attacks. The resulting models are both accurate and accompanied by explicit statistical explanations for remaining errors.

Core claim

XGBoost achieves high detection accuracy on UAVIDS-2025, and SHAP values combined with Westfall-Young tests and Jensen-Shannon distances on kernel density estimates demonstrate that false predictions for Wormhole and Blackhole attacks arise from density support intersections that allow these attacks to mimic normal traffic distributions.

What carries the argument

Shapley Additive explanations (SHAP) for feature attribution together with Westfall-Young permutation tests and Jensen-Shannon distances applied to bandwidth-optimized kernel density estimates for quantifying distribution overlap.

If this is right

Tree-ensemble models like XGBoost outperform hybrid and DNN approaches for this UAV intrusion detection task.
SHAP reveals attack-specific feature manipulations that enable traffic mimicry by Wormhole and Blackhole attacks.
Density support intersection is identified as the mechanism behind persistent misclassifications in two attack types.
The statistical pipeline using permutation tests and Jensen-Shannon distances offers a repeatable method to diagnose overlap-driven errors in other intrusion datasets.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same density-overlap diagnosis could be applied to other network security datasets where attacks are crafted to blend with normal traffic.
Real-time UAV monitoring systems might track the SHAP-highlighted features to flag potential mimicry before full classification.
Future UAVIDS collections could reduce misclassifications by generating attack samples with deliberately lower density support overlap.

Load-bearing premise

The UAVIDS-2025 dataset mirrors real-world UAV network traffic distributions and the statistical tests correctly attribute misclassifications to density overlaps instead of artifacts from modeling or preprocessing.

What would settle it

Repeating the kernel density estimation, violin plot comparison, and Westfall-Young permutation analysis on an independent UAV traffic trace that includes confirmed Wormhole and Blackhole attacks to check whether the same feature overlaps and error patterns reappear.

Figures

Figures reproduced from arXiv: 2605.13922 by Christos Douligeris, Iakovos-Christos Zarkadis.

**Figure 3.** Figure 3: Correlation Coefficients after Dropping Correlated [PITH_FULL_IMAGE:figures/full_fig_p002_3.png] view at source ↗

**Figure 1.** Figure 1: RFE for the Random Forest From the feature selection, the ones selected can be seen in [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

**Figure 4.** Figure 4: RealMLP Stratified 10-Fold CV. III. EXPLAINABILITY To gain actual and trustworthy insight from our trained models, we will perform SHAP analysis, for our best model, XGBoost. Our Methodology starts by examining global and local explanations. Then we move to distribution analysis with violin plots, kernel density estimations and statistical divergence computation. Last but not least, we target the features … view at source ↗

**Figure 5.** Figure 5: XGBoost Confusion Matrices [PITH_FULL_IMAGE:figures/full_fig_p003_5.png] view at source ↗

**Figure 8.** Figure 8: XGBoost Global, Local, Semi-Local Feature Impor [PITH_FULL_IMAGE:figures/full_fig_p004_8.png] view at source ↗

**Figure 6.** Figure 6: XGBoost Global Feature Importances per Label. [PITH_FULL_IMAGE:figures/full_fig_p004_6.png] view at source ↗

**Figure 7.** Figure 7: MLP Global Feature Importances per Label. [PITH_FULL_IMAGE:figures/full_fig_p004_7.png] view at source ↗

**Figure 10.** Figure 10: Feature Box-Plots per Label. make the difference in XGBoost’s decisions. For the same attack, it is also interesting, that in the variable ”RxByteRate/s”, we have observations that aid both in the increase and decrease of the probability, but then it’s heavy right tail slightly thickens at the end, which indicates a second peak in it’s probability density function. The same we observe in the Wormhole att… view at source ↗

**Figure 11.** Figure 11: XGBoost Feature Violin Plots per Label. observed in ”LostPackets”, stand for values below and close to the median of the distribution, after the transformation with the Robust scaler, which are indications of Blackhole attack. This because from both the Box-Plots and Violin plots in [PITH_FULL_IMAGE:figures/full_fig_p005_11.png] view at source ↗

**Figure 14.** Figure 14: Local Explanation for the 7th observation [PITH_FULL_IMAGE:figures/full_fig_p006_14.png] view at source ↗

**Figure 15.** Figure 15: Local Explanation for the 76th observation D. Density Estimation and Westfall-Young Permutation Test To gain deeper statistical insights about our model’s decision’s we will perform the Westfall-Young Permutation Test, for a total of 1000 permutations [16], [17] and Kernel Density Estimation with Bandwidth optimization [25]. Due to the misclassifications that occurred, in the confusion matrix, between t… view at source ↗

**Figure 13.** Figure 13: KDEs per Label [PITH_FULL_IMAGE:figures/full_fig_p006_13.png] view at source ↗

**Figure 16.** Figure 16: Feature: PacketDropRate [PITH_FULL_IMAGE:figures/full_fig_p007_16.png] view at source ↗

**Figure 17.** Figure 17: Feature: LostPackets IV. CONCLUSIONS In this paper we applied best practices for data preprocessing and provided top-performing algorithms and hybrid models, that are capable to detect intrusions in UAV systems, compared to baseline models. After the selection of our best model, we move to SHAP analysis, where we establish solid proof of our model’s internal decision mechanisms, with local and global exp… view at source ↗

**Figure 18.** Figure 18: Feature: AverageHopCount [PITH_FULL_IMAGE:figures/full_fig_p007_18.png] view at source ↗

**Figure 19.** Figure 19: Feature: MeanDelay/s [PITH_FULL_IMAGE:figures/full_fig_p007_19.png] view at source ↗

**Figure 20.** Figure 20: Feature: MeanJitter/s of weak training or data pre-processing, but because of the massive support intersection, in UAVIDS-2025. Even with this major problem our best model has an outstanding performance. By combining all the above techniques, we have a solid view of the attacks. Future work includes cross-dataset validation, with other well-known UAVIDS datasets, that share the same feature space, use of … view at source ↗

read the original abstract

During the last few years, the term Mechanistic Interpretability, a specific area, under the umbrella of explainable artificial intelligence (XAI), has been introduced, to explain the decisions made by complex machine learning (ML) models in critical systems like UAV intrusion detection systems (UAVIDS). In this paper, we apply best-practices for data pre-processing and examine a wide range of tree-ensembles, deep neural networks, hybrid stacking models and the latest ensemble neural networks to detect intrusions in UAV, with stratified 10-fold cross validation. With our top-performing model, XGBoost, we proceed to Shapley Additive explanations (SHAP), to analyze the global and local feature importances and understand which features, each attack targets, to mimic normal traffic and where the misclassifications occur. Furthermore a distribution analysis follows, by visually comparing violin plots and the curves of kernel density estimations. With the Westfall-Young permutation test for multiple comparisons, the Bandwidth optimization of the KDEs and the selection of Jensen-Shannon Distance for the test, we discover the true causes of false predictions, observed in Wormhole and Blackhole attacks in UAVIDS-2025. The findings provide robust, reliable and explainable models for UAV intrusion detection, along with statistical insights, which capture and clarify the masked nature of the attacks, regarding the challenge of Density Support Intersection, between these attacks, in this dataset.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

New UAVIDS-2025 dataset plus standard SHAP and permutation tests on misclassifications, but the density-overlap claim needs ablations to rule out pipeline artifacts.

read the letter

The paper's real contribution is the UAVIDS-2025 dataset and a full empirical pipeline that trains tree ensembles, DNNs, hybrids, and tabular models under stratified 10-fold CV, then applies SHAP to the best performer (XGBoost) before running Westfall-Young tests and Jensen-Shannon distances on optimized KDEs. That sequence is executed cleanly and gives practitioners concrete feature-importance maps plus visual distribution comparisons for Wormhole and Blackhole attacks. The SHAP global and local plots are straightforward to read and the violin-plot plus KDE step is a logical follow-on for spotting overlap regions. Credit for shipping the dataset and the reproducible cross-validation setup; those pieces are useful on their own. The soft spot is the central attribution step. The authors conclude that density support intersection is the true cause of the false predictions, yet the manuscript does not show that the same intersections and significance levels survive when the same tests are run on the hybrid DNN or the other ensembles, or under alternative normalizations and bandwidth choices. Without those checks it remains possible that the detected overlap is partly an artifact of the chosen preprocessing or the tree-ensemble inductive bias rather than a pure data property. The work is therefore best read as a careful case study on one dataset rather than a general demonstration that the statistical procedure isolates the root cause. Readers working on UAV or drone security systems will find the dataset and the SHAP results worth pulling, especially if they need explainable baselines. The paper does not advance new theory or algorithms, so I would not cite it in core methodological work, but the empirical package is solid enough that a serious editor should send it for review with a request for the missing ablations. It is coherent on its own terms and engages the literature honestly.

Referee Report

1 major / 2 minor

Summary. The paper evaluates a range of tree-ensemble, DNN, hybrid stacking, and ensemble neural network models for intrusion detection on the UAVIDS-2025 dataset using stratified 10-fold cross-validation. XGBoost is identified as the top performer; SHAP is then applied to analyze global and local feature importances and locate misclassifications. A subsequent distribution analysis employs violin plots, optimized KDEs, Westfall-Young permutation testing, and Jensen-Shannon distance to attribute false predictions on Wormhole and Blackhole attacks to density support intersections in the feature space.

Significance. If the statistical attribution is shown to be robust to modeling and preprocessing choices, the work supplies both competitive detection models and concrete, falsifiable insights into why certain UAV attacks evade detection, which could guide feature engineering and model design in safety-critical IDS settings.

major comments (1)

[Statistical analysis] Statistical analysis section (post-SHAP paragraph): the claim that Westfall-Young permutation testing plus Jensen-Shannon distance on bandwidth-optimized KDEs isolates density support intersection as the root cause of misclassifications for Wormhole and Blackhole attacks is load-bearing, yet the manuscript reports these tests only on the XGBoost pipeline. No ablation across the hybrid DNN or tabular ensembles also evaluated in the paper, nor across alternative normalizations, is provided to demonstrate invariance of the detected intersections and p-values to model inductive bias or preprocessing artifacts.

minor comments (2)

[Abstract] Abstract: the phrase 'the latest ensemble neural networks' is undefined; the main text should list the exact architectures and hyperparameter ranges used for all families.
[Methods] The manuscript should report the exact number of features retained after preprocessing and any feature-engineering steps that could affect the KDE support analysis.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their thorough review and for highlighting the importance of verifying the robustness of our statistical attribution. We address the major comment below and will incorporate the requested extensions in the revised manuscript.

read point-by-point responses

Referee: [Statistical analysis] Statistical analysis section (post-SHAP paragraph): the claim that Westfall-Young permutation testing plus Jensen-Shannon distance on bandwidth-optimized KDEs isolates density support intersection as the root cause of misclassifications for Wormhole and Blackhole attacks is load-bearing, yet the manuscript reports these tests only on the XGBoost pipeline. No ablation across the hybrid DNN or tabular ensembles also evaluated in the paper, nor across alternative normalizations, is provided to demonstrate invariance of the detected intersections and p-values to model inductive bias or preprocessing artifacts.

Authors: We agree that the load-bearing nature of the statistical claim warrants explicit checks for invariance. In the revised version we will extend the Westfall-Young permutation testing, bandwidth-optimized KDE estimation, and Jensen-Shannon distance calculations to the hybrid stacking and tabular ensemble models already evaluated in the paper. These additional results will be presented alongside the original XGBoost findings to confirm that the density-support intersections for Wormhole and Blackhole attacks persist across model families. We will also report a brief sensitivity check under an alternative normalization scheme to address potential preprocessing artifacts. This extension directly strengthens the falsifiability of the mechanistic insight without altering the core conclusions. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper trains standard tree ensembles and DNNs on the UAVIDS-2025 dataset using stratified cross-validation, selects XGBoost as top performer, applies SHAP for post-hoc feature attribution on model outputs, and then conducts separate distribution analysis via violin plots, optimized KDEs, Westfall-Young permutation testing, and Jensen-Shannon distance on the resulting feature values. None of these steps define a quantity in terms of itself, rename a fitted parameter as a prediction, or rely on self-citation chains for load-bearing uniqueness claims. The statistical attribution of misclassifications to density support intersection is derived from external, falsifiable methods applied to the data and model outputs rather than reducing to the inputs by construction.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The central claims rest on the representativeness of the UAVIDS-2025 dataset and standard ML assumptions about feature distributions and model behavior.

free parameters (2)

XGBoost and DNN hyperparameters
Fitted during cross-validation to achieve reported performance.
KDE bandwidth
Optimized as part of the distribution analysis.

axioms (2)

domain assumption UAVIDS-2025 contains representative samples of normal and attack traffic.
Required for the validity of detection performance and misclassification analysis.
domain assumption Jensen-Shannon distance on KDEs correctly measures overlap causing misclassifications.
Invoked to conclude true causes of false predictions.

pith-pipeline@v0.9.0 · 5575 in / 1282 out tokens · 45456 ms · 2026-05-15T05:45:30.706945+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

29 extracted references · 29 canonical work pages · 2 internal anchors

[1]

Machine Learning Based In- trusion Detection System

A. A. Halimaa, K. Sundarakantham, “Machine Learning Based In- trusion Detection System”, IEEE International Conference on Trends in Electronics and Informatics, October, 2019, pp. 916-920, doi: 10.1109/ICOEI.2019.8862784

work page doi:10.1109/icoei.2019.8862784 2019
[2]

A Taxonomy of Machine-Learning-Based Intrusion Detection Systems for the Internet of Things: A Survey

A. Jamalipour, S. Murali, “A Taxonomy of Machine-Learning-Based Intrusion Detection Systems for the Internet of Things: A Survey”, IEEE Internet of Things Journal, vol. 9, no. 12, November, 2021, pp.9444–9466, doi: 10.1109/JIOT.2021.3126811

work page doi:10.1109/jiot.2021.3126811 2021
[3]

Securing UA V Swarms with Vision Transform- ers: A Byzantine-Robust Federated Learning Framework for Cross- Modal Intrusion Detection

C. B. S ¸ahin, “Securing UA V Swarms with Vision Transform- ers: A Byzantine-Robust Federated Learning Framework for Cross- Modal Intrusion Detection”, Drones, vol.10, no.2, February, doi: 2026https://doi.org/10.3390/drones10020125

work page doi:10.3390/drones10020125
[4]

Better by default: Strong Pre-Tuned MLPs and Boosted Trees on Tabular Data

D. Holzm ¨uller, L. Grinsztajn, I. Steinwart, “Better by default: Strong Pre-Tuned MLPs and Boosted Trees on Tabular Data”, NeurIPS, 2024

work page 2024
[5]

Towards an Interpretable AI Framework for Advanced Classification of Unmanned Aerial Vehicles (UA Vs)

E. Haque, K. Hasan, I. Ahmed, M. S. Alam and T. Islam, “Towards an Interpretable AI Framework for Advanced Classification of Unmanned Aerial Vehicles (UA Vs)”, IEEE 21st Consumer Communications and Networking Conference (CCNC), Las Vegas, NV , USA, 2024, pp. 644- 645, doi: 10.1109/CCNC51664.2024.10454862

work page doi:10.1109/ccnc51664.2024.10454862 2024
[6]

Ensemble-IDS: An En- semble Learning Framework for Enhancing AI-Based Network Intrusion Detection Tasks

I. Bibers, O. Arreche, W. Alayed, M. Abdallah, “Ensemble-IDS: An En- semble Learning Framework for Enhancing AI-Based Network Intrusion Detection Tasks”, Applied Sciences, vo. 15, no. 19, September, 2025, doi: https://doi.org/10.3390/app151910579

work page doi:10.3390/app151910579 2025
[7]

Exact and Approximate Stepdown Methods for Multiple Hypothesis Testing

J. Romano, M. Wolf, “Exact and Approximate Stepdown Methods for Multiple Hypothesis Testing”, SSRN Electronic Journal, February, 2005, doi: 10.2139/ssrn.563267

work page doi:10.2139/ssrn.563267 2005
[8]

Interpretability in Intelligent Systems — A New Concept?

L. K. Hansen, L. Rieger, “Interpretability in Intelligent Systems — A New Concept?”, Springer, September, 2019, pp. 41–49, doi: https://doi.org/10.1007/978-3-030-28954-6

work page doi:10.1007/978-3-030-28954-6 2019
[9]

UA V Airborne Network Intrusion Detection Method Based on Improved Stratified Sampling and Ensemble Learning

L. Lin, H. Ge, Y . Zhou, R. Shangguan, “UA V Airborne Network Intrusion Detection Method Based on Improved Stratified Sampling and Ensemble Learning”, Drones, vol. 9, no. 9, August, 2025, https://doi.org/10.3390/drones9090604

work page doi:10.3390/drones9090604 2025
[10]

Probabilistic Machine Learning: Advanced Topics

K. P. Murphy, “Probabilistic Machine Learning: Advanced Topics”, Cambridge, Massachusetts, USA, MIT Press, 2023, pp.55-236

work page 2023
[11]

A Comparative Analysis of Ensemble-Based Machine Learning Approaches With Explainable AI for Multi-Class Intrusion Detection in Drone Net- works

M. A. Hossain, W. Ishtiaq, and M. S. Islam, “A Comparative Analysis of Ensemble-Based Machine Learning Approaches With Explainable AI for Multi-Class Intrusion Detection in Drone Net- works”, Security and Privacy, vol. 9, no. 1, December, 2025, doi: https://doi.org/10.1002/spy2.70164

work page doi:10.1002/spy2.70164 2025
[12]

Advanced explainable ensemble models for multi-class intrusion detection in heterogeneous drone and industrial networks

M. Islam, F. Ahmed, W. Ishtiaq, M. Hossai, M. Tarek, “Advanced explainable ensemble models for multi-class intrusion detection in heterogeneous drone and industrial networks”, Journal of Information Security, Springer, March, 2026, doi: https://doi.org/10.1186/s13635- 026-00234-w

work page doi:10.1186/s13635- 2026
[13]

AI-Enhanced Intrusion Detection for UA V Systems: A Taxonomy and Comparative Review

M. S. Islam, A. S. Mahmoud, T. R. Sheltami, “AI-Enhanced Intrusion Detection for UA V Systems: A Taxonomy and Comparative Review”, Drones, 2025, vol. 9, no. 10, doi: https://doi.org/10.3390/drones9100682

work page doi:10.3390/drones9100682 2025
[14]

An Explainable Machine Learning-Based Network Intrusion Detection System for Enabling Generalisability in Securing IoT Networks

M. Sarhan, S. Layeghy, M. Portmann, “An Explainable Machine Learning-Based Network Intrusion Detection System for Enabling Generalisability in Securing IoT Networks”, arXiv, April, 2021, doi: https://doi.org/10.48550/arXiv.2104.07183

work page doi:10.48550/arxiv.2104.07183 2021
[15]

AutoGluon-Tabular: Robust and Accurate AutoML for Structured Data

N. Erickson, J. Mueller, A. Shirkov, H. Zhang, P. Larroy, M. Li, A. Smola, “AutoGluon-Tabular: Robust and Accurate AutoML for Structured Data”, arXiv, March, 2020, doi: https://doi.org/10.48550/arXiv.2003.06505

work page internal anchor Pith review doi:10.48550/arxiv.2003.06505 2020
[16]

Asymptotic optimality of the Westfall–Young permutation procedure for multiple testing under dependence

N. Meinshausen, M. H. Maathuis, P. B ¨uhlmann, “Asymptotic optimality of the Westfall–Young permutation procedure for multiple testing under dependence”, The Annals of Statistics, vol. 39, no. 6, December, 2011, pp. 3369–3391, doi: 10.1214/11-AOS946

work page doi:10.1214/11-aos946 2011
[17]

Resampling-Based Multiple Testing: Exam- ples and Methods for p-Value Adjustment

P. H. Westfall, S. S. Young, “Resampling-Based Multiple Testing: Exam- ples and Methods for p-Value Adjustment”, The Annals of Statistics, vol. 43, no. 2, June, 1994, pp. 347–348, doi: https://doi.org/10.2307/2348369

work page doi:10.2307/2348369 1994
[18]

UA VIDS-2025: A Bench- mark Dataset for Intrusion Detection in UA V Networks Using Machine Learning Techniques

Q. Zeng, A. Bashir and F. Nait-Abdesselam, “UA VIDS-2025: A Bench- mark Dataset for Intrusion Detection in UA V Networks Using Machine Learning Techniques”, 2025 IEEE Conference on Communications and Network Security (CNS), Avignon, France, 2025, pp. 1-9, doi: 10.1109/CNS66487.2025.11194990

work page doi:10.1109/cns66487.2025.11194990 2025
[19]

Unmanned aerial vehicles (UA Vs): practical aspects, applications, open challenges, security issues, and future trends

S. A. H. Mohsan, N. Q. H. Othman, Y . Li, M. H. Alsharif, and M. A. Khan, “Unmanned aerial vehicles (UA Vs): practical aspects, applications, open challenges, security issues, and future trends”, In- telligent Service Robotics, vol. 16, no. 1, 2023, pp. 109–137, doi: 10.1007/s11370–022-00452-4

work page doi:10.1007/s11370 2023
[20]

Explaining XgBoost Predictions with SHAP Value: A Com- prehensive Guide to Interpreting Decision Tree-Based Models

S. Erg ¨un, “Explaining XgBoost Predictions with SHAP Value: A Com- prehensive Guide to Interpreting Decision Tree-Based Models”, New Trends in Computer Sciences, vol.1, no.1, April, 2023, pp. 19–31, doi: https://doi.org/10.3846/ntcs.2023.17901

work page doi:10.3846/ntcs.2023.17901 2023
[21]

PnPXAI: A Universal XAI Framework Providing Automatic Explanations Across Diverse Modalities and Models

S. Kim, S. A. Kim, G. Kim, E. Menadjiev, C. Lee, S. Chung, N. Kim, J. Choi, “PnPXAI: A Universal XAI Framework Providing Automatic Explanations Across Diverse Modalities and Models”, arXiv, May, 2025, doi: https://doi.org/10.48550/arXiv.2505.10515

work page doi:10.48550/arxiv.2505.10515 2025
[22]

A Unified Approach to Interpreting Model Predictions

S. M. Lundberg, S. L. Lee, “A Unified Approach to Interpreting Model Predictions”, NeurIPS, December, 2017, pp. 4765–4774, doi: https://doi.org/10.48550/arXiv.1705.07874

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1705.07874 2017
[23]

Explainable Intrusion Detection Sys- tems (X-IDS): A Survey of Current Methods, Challenges, and Opportunities

S. Neupane, J. Ables, W. Anderson, S. Mittal, S. Rahimi, I. Banicescu and M. Seale, “Explainable Intrusion Detection Sys- tems (X-IDS): A Survey of Current Methods, Challenges, and Opportunities”, IEEE Access, vol.10, no.7, January, 2022, doi: https://doi.org/10.48550/arXiv.2207.06236

work page doi:10.48550/arxiv.2207.06236 2022
[24]

Deep Learning Based Intrusion Detection for Cybersecurity in Unmanned Aerial Vehicles Network

S. Niyonsaba, K. Konate and M. M. Soidridine, “Deep Learning Based Intrusion Detection for Cybersecurity in Unmanned Aerial Vehicles Network”, International Conference on Electrical, Computer and En- ergy Technologies (ICECET), Sydney, Australia, 2024, pp. 1-6, doi: 10.1109/ICECET61485.2024.10698453

work page doi:10.1109/icecet61485.2024.10698453 2024
[25]

From Gaussian kernel den- sity estimation to kernel methods

S. Wang, Z. Deng, FL. Chung, W. Hu, “From Gaussian kernel den- sity estimation to kernel methods”. International Journal of Machine Learning and Cybernetics, vol. 4, no. 2, April, 2013, pp.119-37, doi: https://doi.org/10.1007/s13042-012-0078-8

work page doi:10.1007/s13042-012-0078-8 2013
[26]

Agentic Explainable Artificial Intelligence (Agentic XAI) Approach To Explore Better Explanation

T. Yamaguchi, Y . Zhou, M. Ryo, K. Katsura, “Agentic Explainable Artificial Intelligence (Agentic XAI) Approach To Explore Better Explanation”, arXiv, December, 2025, doi: https://doi.org/10.48550/arXiv.2512.21066

work page doi:10.48550/arxiv.2512.21066 2025
[27]

One explanation Does Not Ft All: A Toolkit and Taxonomy of AI Explainability Techniques

V . Arya, R. K. E. Bellamy, P.-Y . Chen, A. Dhurandhar, M. Hind,S. C. Hoffman, S. Houde, Q. V . Liao, R. Luss, A. Mojsilovi ´c, S. Mourad, P. Pedemonte, R. Raghavendra, J. Richards, P. Sattigeri, K. Shanmugam, M. Singh, K. R. Varshney, D. Wei, Y . Zhang, “One explanation Does Not Ft All: A Toolkit and Taxonomy of AI Explainability Techniques” arXiv, Septe...

work page doi:10.48550/arxiv.1909.03012 2019
[28]

DroneGuard: An Explainable and Efficient Machine Learning Framework for Intrusion Detection in Drone Networks

V . U. Ihekoronye, S. O. Ajakwe, J. M. Lee and D.-S. Kim, “DroneGuard: An Explainable and Efficient Machine Learning Framework for Intrusion Detection in Drone Networks”, IEEE Internet of Things Journal, vol. 12, no. 7, pp. 7708-7722, April, 2025, doi: 10.1109/JIOT.2024.3519633

work page doi:10.1109/jiot.2024.3519633 2025
[29]

Multiple Intrusion Detection Using Shapley Additive Explanations and a Heterogeneous Ensemble Model in an Un- manned Aerial Vehicle’s Controller Area Network

Y .-W. Hong, D.-Y . Yoo, “Multiple Intrusion Detection Using Shapley Additive Explanations and a Heterogeneous Ensemble Model in an Un- manned Aerial Vehicle’s Controller Area Network”, Applied Sciences, vol. 14, no. 13, June, 2024, doi: https://doi.org/10.3390/app14135487

work page doi:10.3390/app14135487 2024