pith. machine review for the scientific record. sign in

arxiv: 2604.05790 · v1 · submitted 2026-04-07 · 💻 cs.HC

Recognition: no theorem link

Improving Explanations: Applying the Feature Understandability Scale for Cost-Sensitive Feature Selection

Nicola Rossberg , Bennett Kleinberg , Barry O'Sullivan , Luca Longo , Andrea Visentin
Authors on Pith no claims yet

Pith reviewed 2026-05-10 18:33 UTC · model grok-4.3

classification 💻 cs.HC
keywords explainable AIfeature selectionunderstandabilitymachine learningtabular datamodel interpretabilityco-optimization
0
0 comments X

The pith

Machine learning explanations become more understandable by selecting features according to user comprehension scores while preserving high accuracy.

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

This paper tests whether feature selection guided by the Feature Understandability Scale can improve the quality of natural-language explanations for tabular data models. The authors collect understandability scores for features in two datasets and introduce a co-optimization method that balances these scores against classification accuracy. They show that the resulting models keep strong predictive performance while the chosen features produce explanations that read as more accessible. The work treats interpretability as something that can be designed into the model rather than added afterward. Readers would care because clearer explanations could help people actually use and trust AI decisions in practice.

Core claim

The paper establishes that accuracy and understandability can be successfully co-optimised while maintaining high classification performances. By treating understandability scores as costs in feature selection, the method produces explanations that are considered more understandable at face value on the tested datasets. This is presented as a proof-of-concept contribution to building model interpretability by design.

What carries the argument

The Feature Understandability Scale applied within a co-optimisation methodology for cost-sensitive feature selection.

If this is right

  • Explanations for tabular classification tasks can be made more accessible without sacrificing predictive performance.
  • Feature selection can directly incorporate user comprehension as a design criterion.
  • The co-optimization approach works across at least two different datasets while keeping high accuracy.
  • Interpretability becomes an integrated part of model construction rather than a post-hoc fix.

Where Pith is reading between the lines

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

  • The method could be tested on non-tabular data or other model types to see if the co-optimization still holds.
  • Individual differences in user background might require personalized understandability scores rather than a single scale.
  • Similar cost-based selection could be explored for other explanation qualities such as completeness or causal clarity.

Load-bearing premise

The Feature Understandability Scale provides a reliable measure of how well users will actually comprehend the selected features in explanations.

What would settle it

A user study in which participants rate the understandability of explanations from the co-optimized models versus standard models; if the ratings show no improvement or accuracy falls substantially, the central claim would not hold.

Figures

Figures reproduced from arXiv: 2604.05790 by Andrea Visentin, Barry O'Sullivan, Bennett Kleinberg, Luca Longo, Nicola Rossberg.

Figure 1
Figure 1. Figure 1: Workflow of this study. 3.2 Data Understanding The FUS is applied to two public datasets to ensure that any changes in explanation quality are not domain-specific. The first dataset is the ‘Telco Customer Churn’ dataset, which contains data that can be used to predict whether a customer will switch service providers given their family and usage history. The dataset is publicly available 5 . This dataset wi… view at source ↗
Figure 2
Figure 2. Figure 2: Workflow of the Cost-Sensitive feature selection [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Average feature cost and standard deviation for the phone company customer [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Average feature cost and standard deviation for the medical dataset. Note that [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
read the original abstract

With the growing pervasiveness of artificial intelligence, the ability to explain the inferences made by machine learning models has become increasingly important. Numerous techniques for model explainability have been proposed, with natural-language textual explanations among the most widely used approaches. When applied to tabular data, these explanations typically draw on input features to justify a given inference. Consequently, a user's ability to interpret the explanation depends on their understanding of the input features. To quantify this feature-level understanding, Rossberg et al. introduced the Feature Understandability Scale. Building on that work, this proof-of-concept study collects understandability scores across two datasets, proposes a co-optimisation methodology of understandability and accuracy and presents the resulting explanations alongside the model accuracies. This work contributes to the body of knowledge on model interpretability by design. It is found that accuracy and understandability can be successfully co-optimised while maintaining high classification performances. The resulting explanations are considered more understandable at face value. Further research will aim to confirm these findings through user evaluation.

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.

Referee Report

2 major / 2 minor

Summary. This proof-of-concept manuscript applies the Feature Understandability Scale (Rossberg et al.) to cost-sensitive feature selection for natural-language explanations of tabular ML models. It collects scale scores on two datasets, proposes a co-optimization procedure balancing classification accuracy and feature understandability, and presents the resulting explanations together with model accuracies, claiming that accuracy and understandability can be co-optimized while preserving high performance and yielding more understandable explanations at face value. User studies to confirm comprehension gains are deferred to future work.

Significance. If the co-optimization procedure and scale-based selection prove robust, the work could meaningfully advance human-centered XAI by embedding feature-level understandability directly into model design rather than post-hoc explanation. The explicit reuse of an existing, cited scale is a strength that avoids ad-hoc invention and supports cumulative progress in interpretability research.

major comments (2)
  1. [Abstract] Abstract: the claim that 'accuracy and understandability can be successfully co-optimised while maintaining high classification performances' is stated without any quantitative metrics (accuracy values, understandability scores, baselines, error bars, or dataset identifiers), rendering the central empirical finding impossible to evaluate or reproduce from the provided text.
  2. [Methodology / Results] Methodology and Results sections: the co-optimization rests on the Feature Understandability Scale serving as a reliable proxy for user comprehension, yet the manuscript explicitly defers validation of actual comprehension gains to 'further research … through user evaluation.' This assumption is load-bearing for the claim of improved explanations, as no check is described showing that higher scale scores predict downstream user performance or that the selection step preserves explanation fidelity beyond reported accuracies.
minor comments (2)
  1. [Abstract] The abstract uses the vague qualifier 'at face value'; replace with a concrete statement of what was observed (e.g., mean scale scores or qualitative comparison).
  2. Dataset characteristics, sources, and preprocessing steps are not described; add a dedicated subsection or table to support reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback on our proof-of-concept manuscript. We address each major comment below, proposing targeted revisions where appropriate to improve clarity and precision while preserving the scope of the current work.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that 'accuracy and understandability can be successfully co-optimised while maintaining high classification performances' is stated without any quantitative metrics (accuracy values, understandability scores, baselines, error bars, or dataset identifiers), rendering the central empirical finding impossible to evaluate or reproduce from the provided text.

    Authors: We agree that the abstract would benefit from greater specificity to allow immediate evaluation of the central claim. In the revised version, we will incorporate key quantitative results, including the achieved classification accuracies on both datasets, the corresponding Feature Understandability Scale scores before and after co-optimization, and explicit dataset identifiers. This addition will make the empirical finding reproducible from the abstract while respecting length constraints. revision: yes

  2. Referee: [Methodology / Results] Methodology and Results sections: the co-optimization rests on the Feature Understandability Scale serving as a reliable proxy for user comprehension, yet the manuscript explicitly defers validation of actual comprehension gains to 'further research … through user evaluation.' This assumption is load-bearing for the claim of improved explanations, as no check is described showing that higher scale scores predict downstream user performance or that the selection step preserves explanation fidelity beyond reported accuracies.

    Authors: We acknowledge that the Feature Understandability Scale functions as a proxy in the current study and that direct validation of comprehension gains via user studies is explicitly deferred to future work, as stated in the manuscript. The proof-of-concept contribution is limited to demonstrating that the scale scores and model accuracy can be co-optimized while retaining high classification performance; the 'more understandable at face value' phrasing refers directly to the scale scores obtained. We will revise the Methodology and Results sections to more explicitly frame the scale as a proxy, restate the scope of the current claims, and clarify that fidelity is assessed via maintained accuracy (as a necessary but not sufficient indicator). No additional empirical checks on predictive validity of the scale are available at this stage. revision: partial

Circularity Check

0 steps flagged

No significant circularity; co-optimization claim uses prior scale as input without self-referential reduction

full rationale

The paper's chain consists of citing the Feature Understandability Scale from prior work, collecting scores on two datasets, proposing a co-optimization procedure for feature selection, and reporting empirical outcomes on accuracy and scale scores. No equations, fitted parameters renamed as predictions, or self-definitional loops are present. The self-citation introduces a measurement tool treated as an external input rather than deriving the central claim from itself. The paper acknowledges the need for separate user evaluation, confirming the reported results are not presented as closed within this manuscript. This is a standard application of prior methodology with new empirical steps and does not meet criteria for any enumerated circularity pattern.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No full manuscript text was accessible; therefore no free parameters, axioms, or invented entities could be audited from the paper itself.

pith-pipeline@v0.9.0 · 5487 in / 1053 out tokens · 29077 ms · 2026-05-10T18:33:36.450552+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

71 extracted references · 7 canonical work pages

  1. [1]

    In: World Conference on Explainable Artificial Intelligence

    Ahmed, T., Biecek, P., Longo, L.: Latent space interpretation and mechanistic clipping of subject-specific variational autoencoders of eeg topographic maps for artefacts reduction. In: World Conference on Explainable Artificial Intelligence. pp. 327–350. Springer (2025)

    2025

  2. [2]

    IEEE Access10, 107575–107586 (2022)

    Ahmed, T., Longo, L.: Examining the size of the latent space of convolutional variational autoencoders trained with spectral topographic maps of eeg frequency bands. IEEE Access10, 107575–107586 (2022)

    2022

  3. [3]

    In: Pro- ceedings of the AAAI conference on artificial intelligence

    Arik, S.Ö., Pfister, T.: Tabnet: Attentive interpretable tabular learning. In: Pro- ceedings of the AAAI conference on artificial intelligence. vol. 35, pp. 6679–6687 (2021)

    2021

  4. [4]

    Journal of Reliable Intelligent Environments11(1), 1 (2025)

    Assis, A., Dantas, J., Andrade, E.: The performance-interpretability trade-off: A comparative study of machine learning models. Journal of Reliable Intelligent Environments11(1), 1 (2025)

    2025

  5. [5]

    In: 2023 IEEE Latin American Conference on Computational Intelligence (LA-CCI)

    Assis, A., Véras, D., Andrade, E.: Explainable artificial intelligence-an analysis of the trade-offs between performance and explainability. In: 2023 IEEE Latin American Conference on Computational Intelligence (LA-CCI). pp. 1–6. IEEE (2023)

    2023

  6. [6]

    In: International Conference on Information Systems Architecture and Technology

    Bach, M., Werner, A.: Cost-sensitive feature selection for class imbalance problem. In: International Conference on Information Systems Architecture and Technology. pp. 182–194. Springer (2017)

    2017

  7. [7]

    In: Proceedings of the 43rd international ACM SIGIR conference on research and development in information retrieval

    Balog, K., Radlinski, F.: Measuring recommendation explanation quality: The conflicting goals of explanations. In: Proceedings of the 43rd international ACM SIGIR conference on research and development in information retrieval. pp. 329–338 (2020)

    2020

  8. [8]

    Advanced Drug Delivery Reviews202, 115108 (2023)

    Bao, Z., Bufton, J., Hickman, R.J., Aspuru-Guzik, A., Bannigan, P., Allen, C.: Revolutionizing drug formulation development: The increasing impact of machine learning. Advanced Drug Delivery Reviews202, 115108 (2023)

    2023

  9. [9]

    Minds and Machines33(2), 347–377 (2023)

    Baron, S.: Explainable ai and causal understanding: Counterfactual approaches considered. Minds and Machines33(2), 347–377 (2023)

    2023

  10. [10]

    Explanation is all you need in distillation: Mitigating bias and shortcut learning.arXiv preprint arXiv:2407.09788, 2024

    Bassi, P.R., Cavalli, A., Decherchi, S.: Explanation is all you need in distillation: Mitigating bias and shortcut learning. arXiv preprint arXiv:2407.09788 (2024)

    work page arXiv 2024

  11. [11]

    In: Proceedings of the 2022 ACM conference on fairness, accountability, and transparency

    Bell, A., Solano-Kamaiko, I., Nov, O., Stoyanovich, J.: It’s just not that simple: an empirical study of the accuracy-explainability trade-off in machine learning for pub- lic policy. In: Proceedings of the 2022 ACM conference on fairness, accountability, and transparency. pp. 248–266 (2022)

    2022

  12. [12]

    Computers & Operations Research 106, 169–178 (2019)

    Benítez-Peña, S., Blanquero, R., Carrizosa, E., Ramírez-Cobo, P.: Cost-sensitive feature selection for support vector machines. Computers & Operations Research 106, 169–178 (2019)

    2019

  13. [13]

    Mathematical Problems in Engineering2016(1), 8752181 (2016) Application of the Feature Understandability Scale 21

    Bian, J., Peng, X.g., Wang, Y., Zhang, H.: An efficient cost-sensitive feature selection using chaos genetic algorithm for class imbalance problem. Mathematical Problems in Engineering2016(1), 8752181 (2016) Application of the Feature Understandability Scale 21

    2016

  14. [14]

    In: World Conference on Explainable Artificial Intelligence

    Brdnik, S., Colakovic, I., Karakatič, S.: Non-experts’ trust in xai is unreasonably high. In: World Conference on Explainable Artificial Intelligence. pp. 184–197. Springer (2025)

    2025

  15. [15]

    In: Guidotti, R., Schmid, U., Longo, L

    Cahlik,V.,Alves,R.,Kordik,P.:Reasoning-groundednaturallanguageexplanations for language models. In: Guidotti, R., Schmid, U., Longo, L. (eds.) Explainable Artificial Intelligence. pp. 3–18. Springer Nature Switzerland, Cham (2026)

    2026

  16. [16]

    Communication methods and measures12(1), 25–44 (2018)

    Carpenter, S.: Ten steps in scale development and reporting: A guide for researchers. Communication methods and measures12(1), 25–44 (2018)

    2018

  17. [17]

    In: International Cross-Domain Conference for Machine Learning and Knowledge Extraction

    Carrington, A., Fieguth, P., Chen, H.: Measures of model interpretability for model selection. In: International Cross-Domain Conference for Machine Learning and Knowledge Extraction. pp. 329–349. Springer (2018)

    2018

  18. [18]

    arXiv preprint arXiv:2211.05667 , year=

    Chen, Z., Subhash, V., Havasi, M., Pan, W., Doshi-Velez, F.: What makes a good explanation?: A harmonized view of properties of explanations. arXiv preprint arXiv:2211.05667 (2022)

    work page arXiv 2022

  19. [19]

    In: Proceedings of the 2022 AAAI/ACM Conference on AI, Ethics, and Society

    Dai, J., Upadhyay, S., Aivodji, U., Bach, S.H., Lakkaraju, H.: Fairness via expla- nation quality: Evaluating disparities in the quality of post hoc explanations. In: Proceedings of the 2022 AAAI/ACM Conference on AI, Ethics, and Society. pp. 203–214 (2022)

    2022

  20. [20]

    In: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & data mining

    Danilevsky, M., Dhanorkar, S., Li, Y., Popa, L., Qian, K., Xu, A.: Explainability for natural language processing. In: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & data mining. pp. 4033–4034 (2021)

    2021

  21. [21]

    In: Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems

    DeVrio, A., Cheng, M., Egede, L., Olteanu, A., Blodgett, S.L.: A taxonomy of linguistic expressions that contribute to anthropomorphism of language technologies. In: Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. pp. 1–18 (2025)

    2025

  22. [22]

    In: World Conference on Explainable Artificial Intelligence

    Domnich, M., Veski, R.M., Välja, J., Tulver, K., Vicente, R.: Predicting satisfaction of counterfactual explanations from human ratings of explanatory qualities. In: World Conference on Explainable Artificial Intelligence. pp. 210–229. Springer (2025)

    2025

  23. [23]

    K., Ben-David, S

    Dziugaite, G.K., Ben-David, S., Roy, D.M.: Enforcing interpretability and its statistical impacts: Trade-offs between accuracy and interpretability. arXiv preprint arXiv:2010.13764 (2020)

    work page arXiv 2010

  24. [24]

    Pattern Recognition79, 328–339 (2018)

    Gao, W., Hu, L., Zhang, P.: Class-specific mutual information variation for feature selection. Pattern Recognition79, 328–339 (2018)

    2018

  25. [25]

    Machine learning46(1), 389–422 (2002)

    Guyon, I., Weston, J., Barnhill, S., Vapnik, V.: Gene selection for cancer classifica- tion using support vector machines. Machine learning46(1), 389–422 (2002)

    2002

  26. [26]

    part ii: Explanations

    Halpern, J.Y., Pearl, J.: Causes and explanations: A structural-model approach. part ii: Explanations. The British journal for the philosophy of science (2005)

    2005

  27. [27]

    Psychological Bul- letin107(1), 65 (1990)

    Hilton, D.J.: Conversational processes and causal explanation. Psychological Bul- letin107(1), 65 (1990)

    1990

  28. [28]

    Metrics for Explainable AI: Challenges and Prospects

    Hoffman, R.R., Mueller, S.T., Klein, G., Litman, J.: Metrics for explainable ai: Challenges and prospects. arXiv preprint arXiv:1812.04608 (2018)

    work page Pith review arXiv 2018

  29. [29]

    In: World Conference on Explainable Artificial Intelligence

    Höllig, J., Markus, A.F., De Slegte, J., Bagave, P.: Semantic meaningfulness: evaluating counterfactual approaches for real-world plausibility and feasibility. In: World Conference on Explainable Artificial Intelligence. pp. 636–659. Springer (2023) 22 N. Rossberg et al

    2023

  30. [30]

    KI-Künstliche Intelligenz34(2), 193–198 (2020)

    Holzinger, A., Carrington, A., Müller, H.: Measuring the quality of explanations: the system causability scale (scs) comparing human and machine explanations. KI-Künstliche Intelligenz34(2), 193–198 (2020)

    2020

  31. [31]

    Computers in Human Behavior: Artificial Humans p

    Hunsicker, T., König, C.J., Langer, M.: Investigating choices regarding the accuracy- transparency trade-off of ai-based systems across contexts. Computers in Human Behavior: Artificial Humans p. 100216 (2025)

    2025

  32. [32]

    IEEE Transactions on Systems, Man, and Cybernetics: Systems51(3), 1747–1756 (2019)

    Jiang, L., Kong, G., Li, C.: Wrapper framework for test-cost-sensitive feature selection. IEEE Transactions on Systems, Man, and Cybernetics: Systems51(3), 1747–1756 (2019)

    2019

  33. [33]

    Algorithms18(9), 556 (2025)

    Kabir, S., Hossain, M.S., Andersson, K.: A review of explainable artificial intel- ligence from the perspectives of challenges and opportunities. Algorithms18(9), 556 (2025)

    2025

  34. [34]

    Kästner,L.,Crook,B.:Explainingaithroughmechanisticinterpretability.European journal for philosophy of science14(4), 52 (2024)

    2024

  35. [35]

    In: World Conference on Explainable Artificial Intelligence

    Koenen, N., Wright, M.N.: Toward understanding the disagreement problem in neural network feature attribution. In: World Conference on Explainable Artificial Intelligence. pp. 247–269. Springer (2024)

    2024

  36. [36]

    In: World Conference on Explainable Artificial Intelligence

    Kuhl, U., Artelt, A., Hammer, B.: For better or worse: the impact of counterfactual explanations’ directionality on user behavior in xai. In: World Conference on Explainable Artificial Intelligence. pp. 280–300. Springer (2023)

    2023

  37. [37]

    Journal of Research in Personality43(3), 489–493 (2009)

    Kulas, J.T., Stachowski, A.A.: Middle category endorsement in odd-numbered likert response scales: Associated item characteristics, cognitive demands, and preferred meanings. Journal of Research in Personality43(3), 489–493 (2009)

    2009

  38. [38]

    IEEE Transactions on Image Processing27(3), 1323–1335 (2017)

    Liu, M., Xu, C., Luo, Y., Xu, C., Wen, Y., Tao, D.: Cost-sensitive feature selection by optimizing f-measures. IEEE Transactions on Image Processing27(3), 1323–1335 (2017)

    2017

  39. [39]

    Journal of Medical Internet Research27, e73374 (2025)

    Liu, Y., Liu, C., Zheng, J., Xu, C., Wang, D.: Improving explainability and integrability of medical ai to promote health care professional acceptance and use: mixed systematic review. Journal of Medical Internet Research27, e73374 (2025)

    2025

  40. [40]

    Information Fusion106, 102301 (2024)

    Longo, L., Brcic, M., Cabitza, F., Choi, J., Confalonieri, R., Del Ser, J., Guidotti, R., Hayashi, Y., Herrera, F., Holzinger, A., et al.: Explainable artificial intelligence (xai) 2.0: A manifesto of open challenges and interdisciplinary research directions. Information Fusion106, 102301 (2024)

    2024

  41. [41]

    Nature Machine Intelligence2(1), 2522–5839 (2020)

    Lundberg, S.M., Erion, G., Chen, H., DeGrave, A., Prutkin, J.M., Nair, B., Katz, R., Himmelfarb, J., Bansal, N., Lee, S.I.: From local explanations to global under- standing with explainable ai for trees. Nature Machine Intelligence2(1), 2522–5839 (2020)

    2020

  42. [42]

    Engineering Applications of Artificial Intelligence 136, 109035 (2024)

    Ma, X.A., Xu, H., Liu, Y., Zhang, J.Z.: Class-specific feature selection using fuzzy information-theoretic metrics. Engineering Applications of Artificial Intelligence 136, 109035 (2024)

    2024

  43. [43]

    In: Proceedings of the 26th International Conference on Intelligent User Interfaces

    Mohseni, S., Block, J.E., Ragan, E.: Quantitative evaluation of machine learn- ing explanations: A human-grounded benchmark. In: Proceedings of the 26th International Conference on Intelligent User Interfaces. pp. 22–31 (2021)

    2021

  44. [44]

    Materials Today: Proceedings80, 2803–2806 (2023) Application of the Feature Understandability Scale 23

    Pallathadka, H., Mustafa, M., Sanchez, D.T., Sajja, G.S., Gour, S., Naved, M.: Impact of machine learning on management, healthcare and agriculture. Materials Today: Proceedings80, 2803–2806 (2023) Application of the Feature Understandability Scale 23

    2023

  45. [45]

    Journal of Machine Learning Research12, 2825–2830 (2011)

    Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., Duchesnay, E.: Scikit-learn: Machine learning in Python. Journal of Machine Learning Research12, 2825–2830 (2011)

    2011

  46. [46]

    Interna- tional Journal on Artificial Intelligence Tools21(02), 1240005 (2012)

    Phillips-Wren, G.: Ai tools in decision making support systems: a review. Interna- tional Journal on Artificial Intelligence Tools21(02), 1240005 (2012)

    2012

  47. [47]

    AI and Ethics4(3), 691–698 (2024)

    Placani, A.: Anthropomorphism in ai: hype and fallacy. AI and Ethics4(3), 691–698 (2024)

    2024

  48. [48]

    why should i trust you?

    Ribeiro, M.T., Singh, S., Guestrin, C.: " why should i trust you?" explaining the predictions of any classifier. In: Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining. pp. 1135–1144 (2016)

    2016

  49. [49]

    arXiv preprint arXiv:2510.07050 (2025)

    Rossberg, N., Kleinberg, B., O’Sullivan, B., Longo, L., Visentin, A.: The feature understandability scale for human-centred explainable ai: Assessing tabular feature importance. arXiv preprint arXiv:2510.07050 (2025)

    work page arXiv 2025

  50. [50]

    In: World Conference on Explainable Artificial Intelligence

    Scholbeck, C.A., Funk, H., Casalicchio, G.: Algorithm-agnostic feature attributions for clustering. In: World Conference on Explainable Artificial Intelligence. pp. 217–240. Springer (2023)

    2023

  51. [51]

    arXiv preprint arXiv:2210.07126 (2022)

    Schuff, H., Adel, H., Qi, P., Vu, N.T.: Challenges in explanation quality evaluation. arXiv preprint arXiv:2210.07126 (2022)

    work page arXiv 2022

  52. [52]

    In: Proceedings of the IEEE international conference on computer vision

    Selvaraju, R.R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., Batra, D.: Grad- cam: Visual explanations from deep networks via gradient-based localization. In: Proceedings of the IEEE international conference on computer vision. pp. 618–626 (2017)

    2017

  53. [53]

    In: International Conference on Data & Information Sciences

    Shah, R., Pawar, A., Kumar, M.: Enhancing machine learning model using explain- able ai. In: International Conference on Data & Information Sciences. pp. 287–297. Springer (2023)

    2023

  54. [54]

    World Information Technology and Engineering Journal10(07), 3897–3904 (2023)

    Sharifani, K., Amini, M.: Machine learning and deep learning: A review of methods and applications. World Information Technology and Engineering Journal10(07), 3897–3904 (2023)

    2023

  55. [55]

    performance: Bridging the trade-off in deep learning models

    Shuvra, M.K., Gony, M.N., Fatema, K.: Explainability vs. performance: Bridging the trade-off in deep learning models. International Journal of Advanced Research in Computer Science & Technology (IJARCST)7(5), 10931–10941 (2024)

    2024

  56. [56]

    International Journal of Recent Advances in (2024)

    Simuni, G.: Explainable ai in ml: The path to transparency and accountability. International Journal of Recent Advances in (2024)

    2024

  57. [57]

    Journal of Statistical Modeling and Analytics (JOSMA)7(2) (2025)

    Sinha, A., Nayem, H., Kibria, B.G.: Sample size requirements for the central limit theorem for skewed distributions: A simulation study. Journal of Statistical Modeling and Analytics (JOSMA)7(2) (2025)

    2025

  58. [58]

    Review of Philosophy and Psychology2(1), 77–88 (2011)

    Verdejo, V.M., Quesada, D.: Levels of explanation vindicated. Review of Philosophy and Psychology2(1), 77–88 (2011)

    2011

  59. [59]

    Machine Learning and Knowledge Extraction3(3), 615–661 (2021)

    Vilone, G., Longo, L.: Classification of explainable artificial intelligence methods through their output formats. Machine Learning and Knowledge Extraction3(3), 615–661 (2021)

    2021

  60. [60]

    Information Fusion76, 89–106 (2021) 24 N

    Vilone, G., Longo, L.: Notions of explainability and evaluation approaches for explainable artificial intelligence. Information Fusion76, 89–106 (2021) 24 N. Rossberg et al

    2021

  61. [61]

    In: World Conference on Explainable Artificial Intelligence

    Vilone, G., Longo, L.: Development of a human-centred psychometric test for the evaluation of explanations produced by xai methods. In: World Conference on Explainable Artificial Intelligence. pp. 205–232. Springer (2023)

    2023

  62. [62]

    Wang, Z., Huang, C., Yao, X.: A roadmap of explainable artificial intelligence: Explain to whom, when, what and how? ACM Transactions on Autonomous and Adaptive Systems19(4), 1–40 (2024)

    2024

  63. [63]

    IEEE Access (2025)

    Wiratsin, I.o., Ragkhitwetsagul, C.: Effectiveness of explainable artificial intelli- gence (xai) techniques for improving human trust in machine learning models: A systematic literature review. IEEE Access (2025)

    2025

  64. [64]

    In: Proceedings of the 4th international conference on the practical applications of knowledge discovery and data mining

    Wirth, R., Hipp, J.: Crisp-dm: Towards a standard process model for data mining. In: Proceedings of the 4th international conference on the practical applications of knowledge discovery and data mining. vol. 1, pp. 29–39. Manchester (2000)

    2000

  65. [65]

    International Journal of Human-Computer Studies 193, 103376 (2025)

    Xuan, Y., Small, E., Sokol, K., Hettiachchi, D., Sanderson, M.: Comprehension is a double-edged sword: Over-interpreting unspecified information in intelligible machine learning explanations. International Journal of Human-Computer Studies 193, 103376 (2025)

    2025

  66. [66]

    arXiv preprint arXiv:1907.06831 (2019)

    Yang, F., Du, M., Hu, X.: Evaluating explanation without ground truth in inter- pretable machine learning. arXiv preprint arXiv:1907.06831 (2019)

    work page arXiv 1907

  67. [67]

    International Journal of Approximate Reasoning104, 25–37 (2019)

    Zhao, H., Yu, S.: Cost-sensitive feature selection via the l2, 1-norm. International Journal of Approximate Reasoning104, 25–37 (2019)

    2019

  68. [68]

    Electronics10(5), 593 (2021)

    Zhou, J., Gandomi, A.H., Chen, F., Holzinger, A.: Evaluating the quality of machine learning explanations: A survey on methods and metrics. Electronics10(5), 593 (2021)

    2021

  69. [69]

    Knowledge-based systems95, 1–11 (2016)

    Zhou, Q., Zhou, H., Li, T.: Cost-sensitive feature selection using random forest: Selecting low-cost subsets of informative features. Knowledge-based systems95, 1–11 (2016)

    2016

  70. [70]

    Advances in Neural Information Processing Systems36, 57876–57907 (2023)

    Zimmermann, R.S., Klein, T., Brendel, W.: Scale alone does not improve mecha- nistic interpretability in vision models. Advances in Neural Information Processing Systems36, 57876–57907 (2023)

    2023

  71. [71]

    ACM Computing Surveys55(5), 1–31 (2022)

    Zini, J.E., Awad, M.: On the explainability of natural language processing deep models. ACM Computing Surveys55(5), 1–31 (2022)

    2022