Evaluating Explanation Quality in X-IDS Using Feature Alignment Metrics

BooJoong Kang; Erisa Karafili; Mohammed Alquliti

REVIEW 4 major objections 4 minor 28 references

Evaluating Explanation Quality in X-IDS Using Feature Alignment Metrics

T0 review · 4 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read New metrics grade intrusion-detection explanations by domain knowledge.

desk verdict Three set-based alignment metrics, cleanly defined but not new; the useful part is the domain-reference idea, and the empirical claims need the reference sets published before they can be checked. read the letter →

arxiv 2505.08006 v1 pith:HTSNSZFH submitted 2025-05-12 cs.CR

Mohammed Alquliti , Erisa Karafili , BooJoong Kang This is my paper · ORCID

classification cs.CR

keywords explainableAIintrusiondetectionexplanationevaluationfeaturealignmentmetricsSHAPdomain-informedfeaturesMITREATT&CKCICIDS2017

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper proposes three metrics—Feature Alignment Precision (FAP), Feature Alignment Recall (FAR), and Feature Alignment F1 (FAF1)—that score how well an intrusion-detection system's explanation lines up with the features that security domain knowledge says should matter for each attack. The metrics compare the top-k features a post-hoc explainer like SHAP highlights against predefined domain-informed feature sets derived from the ATT&CK and D3FEND knowledge bases. Applied to three models on the CICIDS2017 dataset, the metrics separate the models: the deep models produce explanations in closer agreement with domain knowledge than Random Forest, especially at small k. If the metrics are right, security analysts get an actionable, quantitative way to choose among explainable IDSs and to decide how many features to show.

What carries the argument

The load-bearing object is the domain-informed feature set $F_c$ for each attack class, manually constructed by mapping each CICIDS2017 attack to ATT&CK technique(s), then to D3FEND detect tactics, with enrichment from contextual resources. The metrics are set-containment ratios: FAP = $\frac{|E_x(k) \cap F_c|}{|E_x(k)|}$, FAR = $\frac{|E_x(k) \cap F_c|}{|F_c|}$, and FAF1 the harmonic mean of the two, where $E_x(k)$ is the explainer's ordered top-k features. The reference set $F_c$ is what turns a generic "top features" list into a domain-grounded score; its quality determines the validity of everything downstream.

What would settle it

Replace each $F_c$ with a randomly drawn feature set of the same size and recompute FAP/FAR/FAF1 for the three models; if the same model ordering (deep models above Random Forest) persists under random references, the metrics are capturing model behaviour rather than domain alignment. A more targeted check: the paper's own results show bot and infiltration classes have empty $F_c$ and hence zero FAR at every k; hand-construct reference sets for those two classes and see whether the reported rankings and conclusions change.

Watch

Extended reading notes

Core claim

The central claim is that explanation quality in an X-IDS can be quantified as alignment between the explainer's top-k features and a reference set of domain-informed features. For each attack class $c$, the paper defines $F_c$, a set of features derived from mapping the attack to ATT&CK techniques and D3FEND detect tactics. FAP then measures the fraction of the top-k features that fall in $F_c$, FAR measures the fraction of $F_c$ covered by the top-k, and FAF1 is their harmonic mean; each metric is computed at the instance, class, and dataset level. Evaluated on CICIDS2017 with SHAP explanations from Random Forest, DNN, and CNN-BiLSTM models, the metrics show that the deep models align with the domain reference much better than Random Forest at small k, and that classes differ widely, with bot and infiltration showing zero alignment because their reference sets are empty. The paper argues these scores give analysts a way to pick models and to set the top-k cutoff.

Load-bearing premise

The paper assumes that the manually built feature sets $F_c$ accurately and completely capture the features that should matter for each attack class in the dataset; if those sets are wrong or incomplete, every FAP, FAR, and FAF1 score measures noise rather than explanation quality.

Editorial extensions

If this is right

Analysts can compare competing explainable IDS models on a domain-grounded scale rather than on fidelity or simplicity alone.
The trade-off curves identify a recommended k: the point where FAF1 peaks, or where FAP and FAR cross, tells how many features to show an analyst.
Consistently low scores for an attack class flag that the domain-informed reference set, not only the model, needs revision.
The same metric definitions apply to other datasets and other post-hoc explainers, since they depend only on top-k lists and reference sets.

Reading between the lines

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

One could validate the metrics by user studies: ask analysts whether explanations with higher FAF1 actually speed up accurate triage; the paper does not run this test.
FAP and FAR over k resemble a precision-recall curve, so an area-under-curve summary could rank models with a single number; the paper stops at per-k values.
The empty bot and infiltration sets suggest the construction of $F_c$ is itself a scientific deliverable; the metrics could be inverted to score knowledge-base completeness rather than model quality.
Because the reference sets are hand-built, inter-annotator agreement on $F_c$ would be a natural reliability check the paper does not report.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

The paper defines FAP, FAR, and FAF1, which are precision, recall, and F1 computed over the top-k explanation features versus a reference set. That's not new math—the formulas are textbook set overlap. What's slightly fresh is the packaging: using MITRE ATT&CK and D3FEND to build the reference sets, so the metrics measure alignment with domain knowledge rather than just generic faithfulness. For a security team choosing among X-IDS models, that's a reasonable scorecard idea.

Credit where it's due: the formal definitions are clean, the three evaluation levels (instance, class, dataset) are sensible, and the paper is unusually honest about its own limitations. It explicitly says bot and infiltration have empty reference sets, admits the feature sets may require fine-tuning, and flags that arithmetic mean aggregation is outlier-sensitive. That transparency suggests the authors know where their weak points are.

The soft spots are real and load-bearing. The biggest one: the domain-informed feature sets F_c are never released. There is no mapping table from ATT&CK techniques to D3FEND tactics to CICIDS2017 feature names. The whole empirical evaluation depends on these hand-built sets, and without them every FAP/FAR/FAF1 value in Table 1 and Figures 2–4 is unverifiable. The stress-test note is right that the link from high-level attack behaviors to low-level flow statistics is underdetermined. The paper implicitly concedes this in Section 4.3, but the concession doesn't fix the absence of the artifact.

Next issue: class- and dataset-level metrics are averaged only over correctly predicted instances (Eqs. 2, 5, 8). If models have different accuracy, you are comparing averages over different populations. That's a genuine confound. Also, FAR for empty F_c is undefined, yet reported as zero; the paper never states the convention. And there are no error bars or repeated runs, so the observed DNN-vs-RF differences could be noise.

These are fixable. Publish the feature sets, give the mapping, compute metrics on all instances or justify the restriction, handle empty sets explicitly, and add variance estimates. With those changes, this becomes a legitimate incremental contribution to XAI evaluation for IDS.

My recommendation: send it to peer review, but mark it as major revision. The idea is useful, the writing is clear, and the authors seem honest. A serious referee could push them to make the reference sets available and clean up the evaluation. As it stands, the empirical claim that the metrics 'effectively distinguish' models is not supported—not because the metrics are wrong, but because the reference sets are absent.

Referee Report

4 major / 4 minor

Summary. The paper defines three metrics—Feature Alignment Precision (FAP), Feature Alignment Recall (FAR), and Feature Alignment F1 (FAF1)—that compare the top-k features of a post hoc explanation (SHAP) for an X-IDS against class-specific domain-informed feature sets F_c derived from MITRE ATT&CK and D3FEND. The metrics are defined at instance, class, and dataset levels (Eqs. 1–9) and evaluated on CICIDS2017 with Random Forest, DNN, and CNN-BiLSTM classifiers. The authors report that the deep models' explanations align better with their reference sets, especially at small k, and argue the metrics can select among X-IDSs and guide the choice of the top-k cutoff.

Significance. If the empirical demonstration is sound, the metrics provide a simple, parameter-free, domain-grounded scorecard for explanation content, filling a real gap because most XAI evaluation focuses on fidelity and simplicity rather than domain relevance. The formal definitions are clean, the three-level aggregation is useful, and the metrics involve no fitted constants, with the reference sets external to the models, which mitigates circularity concerns. The central risk is construct validity: every reported score inherits the quality of the manually constructed F_c, and the current paper neither releases nor independently validates these sets, so the load-bearing empirical claims remain unsupported until that is addressed.

major comments (4)

[Section 4.1 (F_c construction)] The empirical claims in Table 1 and Figures 2–4 rest entirely on the class-specific feature sets F_c, but the paper does not provide the mapping table, the resulting feature lists, or any validation that they align with CICIDS2017's feature names. The authors state that the sets were derived by mapping each attack class to MITRE ATT&CK techniques, then to D3FEND detect tactics, then to dataset feature names, but the intermediate link is underdetermined and the mappings are not shown. This is load-bearing because FAP, FAR, and FAF1 are defined as overlap with these sets; if the sets are wrong, incomplete, or unreproducible, the reported scores measure overlap with an arbitrary list. The authors' own admission in Section 4.3 that the sets 'may require additional fine-tuning' and that bot and infiltration classes have empty sets confirms the fragility. Please release the full mapping, justify each feature selection against D3FEND artifacts, and show sensitivity of the results to alternative plausible mappings.
[Section 3.2, Eqs. (2), (5), (8)] The class-level and dataset-level metrics average only over test instances that are 'correctly predicted as the class c', while the text in Section 3.2 earlier defines X_test_c as test instances that share the same label c. This inconsistency changes the evaluation population across models with different accuracy and excludes all false predictions, so comparisons such as 'DNN vs RF at k=5' may reflect differences in which instances survive the correctness filter rather than differences in explanation alignment. Moreover, explanations for incorrect predictions are relevant in practice and are simply dropped. Please define a single fixed evaluation population (for example, all test instances of class c, or separate correct/incorrect analyses) and report accuracies so the reader can assess the population overlap for each model and class.
[Section 4.2 and Eq. (4)] For bot and infiltration, the paper states that the predefined domain-informed feature set contains no features, but the instance-level FAR in Eq. (4) divides by |F_c|, which is then zero, and the FAF1 in Eq. (7) degenerates. Reporting these classes as zero FAR and including them in the dataset-level averages in Eqs. (3), (6), and (9) silently mixes undefined or degenerate values into the aggregate scores and makes the dataset-level numbers class-composition dependent. Please specify how empty reference sets are handled (exclude the class, mark as not applicable, or use a separate convention) and recompute all aggregates accordingly.
[Section 4.3 and Table 1] The claim that the metrics 'effectively differentiate between X-IDSs' is supported only by descriptive point estimates; no confidence intervals, significance tests, or random baseline (for example, overlap of top-k features with a random or frequency-matched feature set) are reported. At several k values the differences between DNN and CNN-BiLSTM are small (for example, FAF1 0.24 vs 0.16 at k=10 and FAP 0.17 vs 0.20 at k=40), so without uncertainty quantification the differentiation claim is overstated. Please add per-class error bars or bootstrapped intervals and a baseline to show that the metrics separate signal from chance.

minor comments (4)

[Figure 4(caption)] In the caption of Figure 4(b), 'FPR' should be 'FAR', since the text and axis labels refer to the FAP–FAR trade-off.
[Section 4.2 and Table 1] There are typos: 'DDoD/DoS' for 'DDoS/DoS' in Section 4.2, and 'T op-k' in the Table 1 header.
[Section 3.3] The sentence 'A high FAP means that most of the features selected by the model belong to the set of domain-informed features Ex' uses the wrong symbol: it should be F_c, not Ex.
[Eq. (3) and related notation] The averaging notation in Eqs. (3), (6), and (9) should make explicit that the sum runs only over classes with non-empty F_c and with at least one evaluated instance; otherwise the dataset-level aggregates are undefined when such classes are included.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: FAP/FAR/FAF1 are direct set-overlap definitions against an external reference, and the acknowledged fragility of the hand-built feature sets is a validity concern, not a circular reduction.

full rationale

The proposed metrics FAP, FAR, and FAF1 (Eqs. 1-9) are defined directly as set overlaps between the explanation's top-k features E_x(k) and a pre-defined domain-informed feature set F_c. There is no fitted parameter, learned mapping, or statistical estimation: the reported values are arithmetic transcriptions of |E_x(k) ∩ F_c| given the externally constructed F_c. The comparison between RF, DNN, and CNN-BiLSTM is therefore not forced by the metric definition; whether one model outperforms another is an empirical contingency of the SHAP explanations and the chosen F_c. The central weakness—that F_c is hand-mapped from MITRE ATT&CK/D3FEND and not released—is a construct-validity and reproducibility concern about the reference standard, not a circular reduction: the paper never fits F_c to the explanations, nor defines F_c in terms of the top-k outputs, nor imports the rankings from the authors' own prior work. The paper explicitly concedes the fragility in Section 4.3, stating that the predefined domain-informed feature sets 'may require additional fine-tuning to align with features in the selected dataset' and observing zero FAR for bot and infiltration classes because their reference sets are empty. Those admissions show the reference is not engineered to produce the reported ranking. No self-citations are load-bearing, and the metrics are not a renamed version of an existing result used to smuggle in a conclusion. The derivation chain is transparent: the metric values are set overlaps, and the interpretation that those overlaps reflect domain knowledge is an external assumption about F_c, not a circular step.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The central evaluation rests on a small set of external inputs: the MITRE/D3FEND-derived feature sets, the CICIDS2017 benchmark, the three model implementations, and SHAP. The feature sets are the only hand-chosen component that directly determines all metric values; they are unpublished. No invented entities are introduced.

free parameters (2)

Domain-informed feature sets F_c = Not disclosed; manually constructed from MITRE ATT&CK/D3FEND
The reference sets used in Eqs. 1-9 are hand-built in Section 4.1 without publication of the actual features or validation, so the metric's interpretation depends on these unverified choices.
top-k cutoff k = 5, 10, 20, 40, and others in figures
k is a user-supplied parameter of the metrics; the paper sweeps it rather than fitting it, but results and conclusions are k-dependent.

assumptions (5)

standard math Standard set theory and arithmetic for precision/recall
The metrics are defined as set intersections and cardinalities (Eqs. 1-9); no further proof is provided.
domain assumption MITRE ATT&CK and D3FEND provide valid ground truth feature relevance for each attack class
Section 4.1 maps attack classes to ATT&CK techniques and D3FEND detect tactics; the paper does not validate these mappings against independent sources.
domain assumption The CICIDS2017 feature names correspond one-to-one with the features named in the knowledge bases
The derived feature sets are compared directly to SHAP output on the dataset; if the naming or semantics differ, alignment scores are unreliable.
domain assumption SHAP feature importances faithfully reflect the model's decision basis
The evaluation assumes SHAP is a valid explanation of the X-IDS models; this is standard in the XAI literature but unverified for these specific models.
ad hoc to paper Correctly predicted instances are a representative subset for evaluating explanation quality
Eqs. 2, 5, and 8 restrict class-level metrics to instances correctly predicted as class c, which may bias the metric toward easy cases.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Evaluating Explanation Quality in X-IDS Using Feature Alignment Metrics." pith.science (2026). https://pith.science/paper/HTSNSZFH

@misc{pith2026250508006,
  author       = {Pith},
  title        = {Pith review of: Evaluating Explanation Quality in X-IDS Using Feature Alignment Metrics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HTSNSZFH}},
  note         = {Machine review of arXiv:2505.08006}
}

read the original abstract

Explainable artificial intelligence (XAI) methods have become increasingly important in the context of explainable intrusion detection systems (X-IDSs) for improving the interpretability and trustworthiness of X-IDSs. However, existing evaluation approaches for XAI focus on model-specific properties such as fidelity and simplicity, and neglect whether the explanation content is meaningful or useful within the application domain. In this paper, we introduce new evaluation metrics measuring the quality of explanations from X-IDSs. The metrics aim at quantifying how well explanations are aligned with predefined feature sets that can be identified from domain-specific knowledge bases. Such alignment with these knowledge bases enables explanations to reflect domain knowledge and enables meaningful and actionable insights for security analysts. In our evaluation, we demonstrate the use of the proposed metrics to evaluate the quality of explanations from X-IDSs. The experimental results show that the proposed metrics can offer meaningful differences in explanation quality across X-IDSs and attack types, and assess how well X-IDS explanations reflect known domain knowledge. The findings of the proposed metrics provide actionable insights for security analysts to improve the interpretability of X-IDS in practical settings.

Figures

Figures reproduced from arXiv: 2505.08006 by the authors.

**Figure 2.** (a) FAR and (b) FAP at various top-k cutoffs for DNN, RF, and CNNBiLSTM X-IDSs. These metrics show how well each X-IDS’s top-k features align with the set of domain-informed features across the entire dataset. Class level explanation evaluation results: We also compute the class level FAP, and FAR by averaging the instance-level scores for each attack class. This provides a detailed view of how well each of the exp… view at source ↗

**Figure 3.** Class level explanation evaluation metrics across attack types for the DNN [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: (a) Class Level FAF1 curve illustrating how each model’s top- [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 26 canonical work pages

[1]

Explainable artificial intelligence in cybersecurity: A survey,

N. Capuano, G. Fenza, V. Loia, and C. Stanzione, “Explainable artificial intelligence in cybersecurity: A survey,”IEEE Access, vol. 10, pp. 93575–93600, 2022

work page 2022
[2]

Explainable intrusion detection for cyber defences in the internet of things: Opportunities and solutions,

N. Moustafa, N. Koroniotis, M. Keshk, A. Y. Zomaya, and Z. Tari, “Explainable intrusion detection for cyber defences in the internet of things: Opportunities and solutions,” IEEE Commun. Surveys Tuts., 2023

work page 2023
[3]

Evaluating feature relevance XAI in network intrusion detection,

J. Tritscher, M. Wolf, A. Hotho, and D. Schlör, “Evaluating feature relevance XAI in network intrusion detection,” inWorld Conf. Explainable Artif. Intell., 2023, pp. 483–497

work page 2023
[4]

Feature relevance XAI in anomaly detection: Reviewing approaches and challenges,

J. Tritscher, A. Krause, and A. Hotho, “Feature relevance XAI in anomaly detection: Reviewing approaches and challenges,”Front. Artif. Intell., vol. 6, 1099521, 2023

work page 2023
[5]

arXiv preprint arXiv:1909.03012 (2019)

Arya, V., Bellamy, R.K., Chen, P.-Y., Dhurandhar, A., Hind, M., Hoffman, S.C., Houde, S., Liao, Q.V., Luss, R., Mojsilović, A., et al.: One explanation does not fit all: A toolkit and taxonomy of AI explainability techniques. arXiv preprint arXiv:1909.03012 (2019)

arXiv 2019
[6]

Lopes, P., Silva, E., Braga, C., Oliveira, T., Rosado, L.: XAI systems evaluation: A review of human and computer-centred methods.Applied Sciences 12(19), 9423 (2022)

work page 2022
[7]

An explainable ensemble deep learning approach for intrusion detection in industrial internet of things,

M. K. Hasan, R. Sulaiman, S. Islam, A. U. Rehman,et al., “An explainable ensemble deep learning approach for intrusion detection in industrial internet of things,”IEEE Access, 2023

work page 2023
[8]

From black boxes to actionable insights: a perspective on explainable artificial intelligence for scientific discovery,

Z. Wu, J. Chen, Y. Li, Y. Deng, H. Zhao, C.-Y. Hsieh, and T. Hou, “From black boxes to actionable insights: a perspective on explainable artificial intelligence for scientific discovery,”J. Chem. Inf. Model., vol. 63, no. 24, pp. 7617–7627, 2023. 18 M. Alquliti et al

work page 2023

Show all 28 references

[9]

https://www.cisa.gov/news-events/news/best- practices-mitre-attckr-mapping, last accessed April 12, 2025

Cybersecurity and Infrastructure Security Agency (CISA): Best practices: MITRE ATT&CK ® mapping. https://www.cisa.gov/news-events/news/best- practices-mitre-attckr-mapping, last accessed April 12, 2025

2025
[10]

In: Proceed- ings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining (KDD 2021), pp

Lin, Y.-S., Lee, W.-C., Celik, Z.B.: What do you see? Evaluation of explainable artificial intelligence (XAI) interpretability through neural backdoors. In: Proceed- ings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining (KDD 2021), pp. 1027–1035 (2021)

2021
[11]

Large Language Models

Nir, D., Kaiser, F.K., Giladi, S., Sharabi, S., Moyal, R., Shpolyansky, S., Murillo, A., Elyashar, A., Puzis, R.: Labeling Network Intrusion Detection System (NIDS) Rules with MITRE ATT&CK Techniques: Machine Learning vs. Large Language Models. Big Data and Cognitive Computing...

2025
[12]

Schwalbe, G., Finzel, B.: A comprehensive taxonomy for explainable artificial in- telligence: A systematic survey of surveys on methods and concepts.Data Mining and Knowledge Discovery38(5), 3043–3101 (2024)

2024
[13]

The MITRE Corporation, vol

Kaloroumakis, P.E., Smith, M.J.: Toward a knowledge graph of cybersecurity coun- termeasures. The MITRE Corporation, vol. 11 (2021)

2021
[14]

Rosenfeld, A.: Better metrics for evaluating explainable artificial intelligence: Blue skyideastrack.In:Proceedingsofthe21thInternationalConferenceonAutonomous Agents and Multiagent Systems (AAMAS 2021), Online, 2021, pp. 3–7

2021
[15]

Hedström, A., Weber, L., Krakowczyk, D., Bareeva, D., Motzkus, F., Samek, W., Lapuschkin, S., Höhne, M.M.-C.: Quantus: An explainable AI toolkit for responsible evaluation of neural network explanations and beyond.Journal of Machine Learning Research 24(34), 1–11 (202X)
[16]

Towards understanding alerts raised by unsupervised network intrusion detection systems,

M. Lanvin, P.-F. Gimenez, Y. Han, F. Majorczyk, L. Mé, and E. Totel, “Towards understanding alerts raised by unsupervised network intrusion detection systems,” in Proc. 26th Int. Symp. Res. Attacks, Intrusions Defenses, 2023, pp. 135–150

2023
[17]

Hedström, A., Weber, L., Krakowczyk, D., Bareeva, D., Motzkus, F., Samek, W., Lapuschkin, S., Höhne, M.M.-C.: Quantus: An explainable AI toolkit for responsible evaluation of neural network explanations and beyond.Journal of Machine Learning Research 24(34), 1–11 (2023)

2023
[18]

Nauta, M., Trienes, J., Pathak, S., Nguyen, E., Peters, M., Schmitt, Y., Schlötterer, J., Van Keulen, M., Seifert, C.: From anecdotal evidence to quantitative evaluation methods: A systematic review on evaluating explainable AI.ACM Computing Sur- veys 55(13s), 1–42 (2023)

2023
[19]

Sharafaldin, I., Lashkari, A.H., Ghorbani, A.A.: Toward generating a new intrusion detection dataset and intrusion traffic characterization.ICISSp 1(2018), 108–116 (2018)

2018
[20]

Applied Sciences 14(10), 4170 (2024)

Arreche, O., Guntur, T., Abdallah, M.: XAI-IDS: Toward proposing an explainable artificial intelligence framework for enhancing network intrusion detection systems. Applied Sciences 14(10), 4170 (2024)

2024
[21]

In:2020 3rd International Conference on Artificial Intelligence and Pattern Recognition, pp

Sinha, J., Manollas, M.: Efficient deep CNN-BiLSTM model for network intru- sion detection. In:2020 3rd International Conference on Artificial Intelligence and Pattern Recognition, pp. 223–231 (2020)

2020
[22]

In: Advances in Neural Information Processing Systems, vol

Lundberg, S.M., Lee, S.-I.: A unified approach to interpreting model predictions. In: Advances in Neural Information Processing Systems, vol. 30 (2017)

2017
[23]

Master’s thesis, Computer Science and Electronic Engineering (CSEE), University of Essex, Colchester, UK (2018)

Kostas, K.: Anomaly Detection in Networks Using Machine Learning. Master’s thesis, Computer Science and Electronic Engineering (CSEE), University of Essex, Colchester, UK (2018)

2018
[24]

https://attack.mitre.org/, last accessed 2025/04/19 Evaluating Explanation Quality in X-IDS 19

MITRE Corporation: MITRE ATT&CK ® Knowledge Base, v16.1 (release 2024-10-31). https://attack.mitre.org/, last accessed 2025/04/19 Evaluating Explanation Quality in X-IDS 19

2024
[25]

arXiv preprint arXiv:2012.00093 (2020)

Dieber, J., Kirrane, S.: Why model why? Assessing the strengths and limitations of LIME. arXiv preprint arXiv:2012.00093 (2020)

2020 arXiv
[26]

Neupane, S., Ables, J., Anderson, W., Mittal, S., Rahimi, S., Banicescu, I., Seale, M.: Explainable intrusion detection systems (X-IDS): A survey of current methods, challenges, and opportunities.IEEE Access 10, 112392–112415 (2022)

2022
[27]

Pawlicki, M., Pawlicka, A., Kozik, R., Choraś, M.: The survey on the dual nature of XAI challenges in intrusion detection and their potential for AI innovation.Artificial Intelligence Review 57(12), 1–32 (2024)

2024
[28]

Zhou, J., Gandomi, A.H., Chen, F., Holzinger, A.: Evaluating the quality of ma- chine learning explanations: A survey on methods and metrics.Electronics 10(5), 593 (2021)

2021

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

Explainable artificial intelligence in cybersecurity: A survey,

N. Capuano, G. Fenza, V. Loia, and C. Stanzione, “Explainable artificial intelligence in cybersecurity: A survey,”IEEE Access, vol. 10, pp. 93575–93600, 2022

work page 2022

[2] [2]

Explainable intrusion detection for cyber defences in the internet of things: Opportunities and solutions,

N. Moustafa, N. Koroniotis, M. Keshk, A. Y. Zomaya, and Z. Tari, “Explainable intrusion detection for cyber defences in the internet of things: Opportunities and solutions,” IEEE Commun. Surveys Tuts., 2023

work page 2023

[3] [3]

Evaluating feature relevance XAI in network intrusion detection,

J. Tritscher, M. Wolf, A. Hotho, and D. Schlör, “Evaluating feature relevance XAI in network intrusion detection,” inWorld Conf. Explainable Artif. Intell., 2023, pp. 483–497

work page 2023

[4] [4]

Feature relevance XAI in anomaly detection: Reviewing approaches and challenges,

J. Tritscher, A. Krause, and A. Hotho, “Feature relevance XAI in anomaly detection: Reviewing approaches and challenges,”Front. Artif. Intell., vol. 6, 1099521, 2023

work page 2023

[5] [5]

arXiv preprint arXiv:1909.03012 (2019)

Arya, V., Bellamy, R.K., Chen, P.-Y., Dhurandhar, A., Hind, M., Hoffman, S.C., Houde, S., Liao, Q.V., Luss, R., Mojsilović, A., et al.: One explanation does not fit all: A toolkit and taxonomy of AI explainability techniques. arXiv preprint arXiv:1909.03012 (2019)

arXiv 2019

[6] [6]

Lopes, P., Silva, E., Braga, C., Oliveira, T., Rosado, L.: XAI systems evaluation: A review of human and computer-centred methods.Applied Sciences 12(19), 9423 (2022)

work page 2022

[7] [7]

An explainable ensemble deep learning approach for intrusion detection in industrial internet of things,

M. K. Hasan, R. Sulaiman, S. Islam, A. U. Rehman,et al., “An explainable ensemble deep learning approach for intrusion detection in industrial internet of things,”IEEE Access, 2023

work page 2023

[8] [8]

From black boxes to actionable insights: a perspective on explainable artificial intelligence for scientific discovery,

Z. Wu, J. Chen, Y. Li, Y. Deng, H. Zhao, C.-Y. Hsieh, and T. Hou, “From black boxes to actionable insights: a perspective on explainable artificial intelligence for scientific discovery,”J. Chem. Inf. Model., vol. 63, no. 24, pp. 7617–7627, 2023. 18 M. Alquliti et al

work page 2023