General Pitfalls of Model-Agnostic Interpretation Methods for Machine Learning Models
Open accessThe DOI matches its Crossref record; 134 of the 134 references it cites were found in a registry or an index.
Abstract
An increasing number of model-agnostic interpretation techniques for machine learning (ML) models such as partial dependence plots (PDP), permutation feature importance (PFI) and Shapley values provide insightful model interpretations, but can lead to wrong conclusions if applied incorrectly. We highlight many general pitfalls of ML model interpretation, such as using interpretation techniques in the wrong context, interpreting models that do not generalize well, ignoring feature dependencies, interactions, uncertainty estimates and issues in high-dimensional settings, or making unjustified causal interpretations, and illustrate them with examples. We focus on pitfalls for global methods that describe the average model behavior, but many pitfalls also apply to local methods that explain individual predictions. Our paper addresses ML practitioners by raising awareness of pitfalls and identifying solutions for correct model interpretation, but also addresses ML researchers by discussing open issues for further research.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures
![Fig. 2 Assuming one-fits-all interpretability . A default xgboost regression model that minimizes the mean squared error (MSE) was fitted on 20 independently and uniformly distributed features to predict another independent, uniformly sampled target. In this setting, predicting the (unconditional) mean E [ Y ] in a constant model is optimal. The learner overfits due to a small training data size. Mean marginal SHAP (red, error bars indicate 0 . 05 and 0 . 95 quantiles) exposes all mechanistically used features. In contrast, PFI on test data (blue, error bars indicate 0 . 05 and 0 . 95 quantiles) considers all features to be irrelevant, since no feature contributes to the generalization performance.](https://figures.cochat.ai/paper-figures/d89fa50113fe4204a130eac0040da274/general-pitfalls-of-model-agnostic-interpretation-methods-for-machine-learning-figure-2.png)


![Fig. 5 Confusing linear correlation with dependence . Highly dependent features X 1 and X 2 that have a correlation close to zero. A test ( H 0 : Features are independent) using Pearson correlation is not significant, but for HSIC, the H 0 -hypothesis gets rejected. Data from [80].](https://figures.cochat.ai/paper-figures/90068dfc976d4ead9a6c7cbb19095454/general-pitfalls-of-model-agnostic-interpretation-methods-for-machine-learning-figure-5.png)

![Fig. 7 Misleading effect due to interactions . Simulation example with interactions: Y = 3 X 1 -6 X 2 + 12 X 2 1 ( X 3 ≥ 0) + ϵ with X 1 , X 2 , X 3 i.i.d. ∼ U [ -1 , 1] and ϵ i.i.d. ∼ N (0 , 0 . 3). A random forest with 500 trees is fitted on 1000 observations. Effects are calculated on 200 randomly sampled (training) observations. A, B: PDP (yellow) and ICE curves of X 1 and X 2 ; C: Derivative ICE curves and their standard deviation of X 2 ; D: 2-dimensional PDP of X 2 and X 3 .](https://figures.cochat.ai/paper-figures/acbe517b1cfe4f9b8d1eb0df227240b1/general-pitfalls-of-model-agnostic-interpretation-methods-for-machine-learning-figure-7.png)
![Fig. 8 Ignoring model and approximation uncertainty . PDP for X 1 with Y = 0 · X 1 + ∑ 10 j =2 X j + ϵ i with X 1 , . . . , X 10 ∼ U [0 , 1] and ϵ i ∼ N (0 , 0 . 9). Left: PDP for X 1 of a random forest trained on 100 data points. Middle: Multiple PDPs (10x) for the model from left plots, but with different samples (each n=100) for PDP estimation. Right: Repeated (10x) data samples of n=100 and newly fitted random forest.](https://figures.cochat.ai/paper-figures/bf95f6cae70f49538e3b3bfd16542e68/general-pitfalls-of-model-agnostic-interpretation-methods-for-machine-learning-figure-8.png)


Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“However, these methods can be misleading when correlations exist between input variables 107,109 .”
From Machine learning in concrete science: applications, challenges, and best practices (2022) · cites this paper as [109]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Breiman, L. (2001) · Machine Learning · doi:10.1023/a:1010933404324 · Cited by 133,538 · cited 1×
“Global methods include the partial dependence plot (PDP) [30], partial importance (PI) [16], accumulated local affects (ALE) [2], or the permutation feature importance (PFI) [10,28,16].”
MethodScikit-learn: Machine Learning in Python
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Müller, A., Nothman, J., Louppe, G., et al. (2012) · arXiv · doi:10.48550/arxiv.1201.0490 · Cited by 71,016 · cited 2×
“These resampling procedures are readily available in software [58,76], and well-studied in theory as well as practice [3,9,88], although rigorous analysis of cross-validation is still considered an open problem [87].”
BackgroundRegression shrinkage and selection via the lasso
Tibshirani, R. (1996) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.2517-6161.1996.tb02080.x · Cited by 51,482 · cited 1×
“An alternative is to directly use models that implicitly perform feature selection such as LASSO [95] or component-wise boosting [84] as they can produce sparse models with fewer features.”
MethodElements of Information Theory
Cover, T. M., & Thomas, J. A. (2001) · Wiley eBooks · doi:10.1002/0471200611 · Cited by 36,808
A Unified Approach to Interpreting Model Predictions
Lundberg, S., & Lee, S.-I. (2017) · arXiv · doi:10.48550/arxiv.1705.07874 · Cited by 23,707 · cited 5×
“The variance of the estimates reduces in O ( 1 m ), where m is the number of evaluated orderings [21,67].”
BackgroundMethodInfluentialA Simple Sequentially Rejective Multiple Test Procedure
(1979) · Scandinavian Journal of Statistics · doi:10.2307/4615733 · Cited by 21,526 · cited 1×
“Since MCP is well known in statistics, we refer the practitioner to [24] for an overview and discussion of alternative adjustment methods, such as the Bonferroni-Holm method [44].”
Method"Why Should I Trust You?": Explaining the Predictions of Any Classifier
Ribeiro, M. T., Singh, S., & Guestrin, C. (2016) · Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining · doi:10.1145/2939672.2939778 · Cited by 17,555
Elements of Information Theory
Cover, T. M., & Thomas, J. A. (2005) · doi:10.1002/047174882x · Cited by 13,791
Gauriat, C.-M., Pencolé, Y., Ribot, P., & Brouillet, G. (2024) · HAL (Le Centre pour la Communication Scientifique Directe) · doi:10.4230/oasics.dx.2024.27 · Cited by 12,586
Rudin, C. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0048-x · Cited by 10,561 · cited 3×
“However, further research is required, as there is no single perfect definition of interpretability, but rather multiple depending on the context [25,82].”
BackgroundMethod
Research works printed but not linked: 6
- Brenning, A.: Transforming feature space to interpret machine learning models. arXiv:2104.04295 (2021) · arXiv
- Freiesleben, T.: Counterfactual explanations & adversarial examples-common grounds, essential differences, and potential transfers. arXiv preprint arXiv:2009.05487 (2020) · arXiv
- Curran Associates, Inc. (2017). https://doi.org/ 10.5555/3295222.3295230 · doi:10.5555/3295222.3295230
- Molnar, C., Freiesleben, T., König, G., Casalicchio, G., Wright, M.N., Bischl, B.: Relating the partial dependence plot and permutation feature importance to the data generating process. arXiv preprint arXiv:2109.01433 (2021) · arXiv
- Semenova, L., Rudin, C., Parr, R.: A study in Rashomon curves and volumes: a new perspective on generalization and model simplicity in machine learning. arXiv preprint arXiv:1908.01755 (2021) 103. · arXiv
- Alzate, C., et al. (eds.) MIDAS/PAP -2018. LNCS (LNAI), vol. 11054, pp. 81-94. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-13463-1 6 · DOI
Cited by
Items 201–300 of 313
Kim, D., Oh, S., Choi, S., & Hong, G. (2026) · Developments in the Built Environment · doi:10.1016/j.dibe.2026.100919 · Cited by 1
Ghale, S., Oka, S., & Takefuji, Y. (2026) · Fuel · doi:10.1016/j.fuel.2026.139411 · Cited by 1
An Interpretable Prediction Method for Tubing Corrosion Based on CASA-XGBoost and SHAP-Sobol
Wu, J., Zhang, Z., Zhao, B., Chen, H., & Wan, L. (2026) · Algorithms · doi:10.3390/a19060430 · Cited by 1
Szeląg, P., POPŁAWSKI, T., & Adamusiński, M. (2026) · Energies · doi:10.3390/en19184426 · Cited by 1
Interpretability of polypharmacy deep-learning models
Dib, L. (2021) · doi:10.21428/594757db.5ed5090e · Cited by 1
Siagian, R. C., Alfaris, L., Nurahman, A., Muhammad, A. C., Nyuswantoro, U. I., & Nasution, B. (2023) · Kappa Journal · doi:10.29408/kpj.v7i2.20921 · Cited by 1
Local Point-wise Explanations of LambdaMART
Rahnama, A. H. A., Bütepage, J., & Boström, H. (2024) · Linköping electronic conference proceedings · doi:10.3384/ecp208014 · Cited by 1
“The logic behind this is that as the sample size grows, these explanations must become consistent since the surrogate model has more information about the vicinity of the document it explains [9].”
BackgroundMethodInfluentialLocal List-Wise Explanations of LambdaMART
Rahnama, A. H. A., Bütepage, J., & Boström, H. (2024) · Communications in computer and information science · doi:10.1007/978-3-031-63797-1_19 · Cited by 1
Fast Explanation of RBF-Kernel SVM Models Using Activation Patterns
Zhang, M., Treder, M. S., Marshall, D., & Li, Y. (2024) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn60899.2024.10650697 · Cited by 2
“It is crucial to consider these factors to ensure the reliability of such methods, as ignoring them may lead to inaccurate results [17].”
Fajar, A., Yazid, S., & Budi, I. D. (2026) · Media Jurnal Informatika · doi:10.35194/mji.v18i1.6501 · Cited by 3
“However, these studies often focus on a single type of model or use qualitative assessments of explainability, lacking a systematic, empirical comparison between black-box and white-box approaches using quantitative metrics [28], [29].”
Advances in AI-Enhanced Biomedical Imaging for Cancer Immunology
Yim, W. W., Wee, F. Y. T., Ho, Z. Y., Feng, X., Zhang, M., Lee, S., Zlobec, I., Yeong, J. P.-S., & Lau, M. C. (2024) · World Scientific Annual Review of Cancer Immunology · doi:10.1142/s2972389224300044 · Cited by 1
fmeffects: An R Package for Forward Marginal Effects
Löwe, H., Scholbeck, C. A., Heumann, C., Bischl, B., & Casalicchio, G. (2025) · The R Journal · doi:10.32614/rj-2024-024 · Cited by 1
Self-citationTakefuji, Y. (2025) · International Immunopharmacology · doi:10.1016/j.intimp.2025.115238 · Cited by 1
Interpretable machine learning for time-to-event prediction in medicine and healthcarePreprint
Baniecki, H., Sobieski, B., Szatkowski, P., Bombinski, P., & Biecek, P. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2303.09817 · Cited by 2
“We foresee several limitations of the introduced methods related to general limitations of post-hoc interpretability methods [55].”
Explainable Machine Learning for Precision Medicine of Patients with Infectious Diseases
Zucco, A. G. (2022) · Research at the University of Copenhagen (University of Copenhagen) · doi:10.13140/rg.2.2.11403.57122 · Cited by 1
Monke, H., Fresz, B., Bernreuther, M., Chen, Y., & Huber, M. F. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2511.09299 · Cited by 1
Popoola, G., & Sheppard, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.09852 · Cited by 1
Models trained with noisy genomes extend bacterial phenotype prediction into deep timePreprint
Koldaeva, A., Bagrova, O., Mitchell, J. A. M., Hugenholtz, P., Spang, A., Woodcroft, B. J., Williams, T. A., & Szöllősi, G. J. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.06.29.735159 · Cited by 1
Pan, C., Meng, A., & Suk, Y. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.21165 · Cited by 1
“When they are correlated, the averaging step forces the model to extrapolate into regions of the covariate space where little or no data exist, and the resulting curves can be badly misleading (Hooker et al., 2021; Molnar et al., 2022, 2023).”
(2021) · Hydrology and earth system sciences · doi:10.5194/hess-2021-9-rc2
(2021) · Hydrology and earth system sciences · doi:10.5194/hess-2021-9-rc1
Peichl, M. (2021) · Hydrology and earth system sciences · doi:10.5194/hess-2021-9-ac1
Algebraic Learning: Towards Interpretable Information ModelingPreprint
Yang, T. O. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2203.06690
Flogard, E. L., Mengshoel, O. J., Theisen, O. M., & Bach, K. (2023) · Frontiers in artificial intelligence and applications · doi:10.3233/faia230644
Machine learning modeling methodology for industrial solid ash
Qi, C., Yilmaz, E., & Chen, Q. (2024) · Elsevier eBooks · doi:10.1016/b978-0-443-15524-6.00013-3
mlr3summary: Concise and interpretable summaries for machine learning modelsPreprint
Dandl, S., Becker, M., Bischl, B., Casalicchio, G., & Bothmann, L. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2404.16899
“Recommended alternatives to permutation feature importances like conditional feature importance [6], are currently not available in a proper R package (published on CRAN).”
Self-citationStatistical interpretation of high-dimensional complex prediction models for biomedical data
Chamma, A. (2024) · theses.fr (ABES) · doi:10.70675/6ea1a1ddz4846z496czbd68zc868183497a9 · Cited by 3
Kempt, H. (2024) · Springer eBooks · doi:10.1007/978-3-031-68098-4_3
Kim, H., & Lim, C. C. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.5040347
Guo, Y. (2025) · International Conference on Big Data, Computational Intelligence, and Applications (BDCIA) · doi:10.1117/12.3059881
“Most focus primarily on parameter adjustment, often neglecting the biological plausibility of the models [15].”
LSTM Ensemble Learning Model with Additive Gaussian Process Priors
Sun, Z., Wang, Y., & Yin, H. (2025) · Highlights in Science Engineering and Technology · doi:10.54097/wfvzd034
“For instance, Molnar et al. [8] have highlighted that the "black box" nature of deep learning models substantially circumscribes their practical application in the economic sphere.”
Jaeger, B. C., Rigdon, J., Weiss, M., Yelton, P., Allen, N., Pajewski, N. M., Tajeu, G. S., Craft, S., Mielke, M. M., & Williamson, J. D. (2025) · medRxiv · doi:10.1101/2025.04.02.25325089
New analytical approaches to life course epidemiology
Börnhorst, C., Wright, M. N., & Didelez, V. (2025) · Oxford University Press eBooks · doi:10.1093/oso/9780198895961.003.0005
Lavelle-Hill, R. E., Smith, G., & Murayama, K. (2023) · OSF Preprints (OSF Preprints) · doi:10.31219/osf.io/6xt82_v1
Momot, О. M., & Grzenda, W. (2025) · Annales Universitatis Mariae Curie-Skłodowska sectio H Oeconomia · doi:10.17951/h.2025.59.1.131-151
Interpretierbares Machine Learning
Hilbert, S., Kraus, E., & Lindl, A. (2025) · Springer eBooks · doi:10.1007/978-3-658-43649-0_6
Linders, G. M., Krapp, L. S., & Schnell, S. (2025) · PsyArXiv (OSF Preprints) · doi:10.31234/osf.io/v4krw_v1
Mathieu, J.-P., Kronwitter, S., Pfahler, J., Roberto, M. D., & Vielhaber, M. (2025) · Proceedings of the Design Society · doi:10.1017/pds.2025.10168 · Cited by 1
“It is important to highlight that what constitutes an interpretable model can vary based on the situation and target user group, making it challenging to determine the level of interpretability required (Molnar et al., 2022).”
InfluentialAkın, A., & Bozçelik, E. E. (2025) · Sosyal Politika Çalışmaları Dergisi · doi:10.21560/spcd.vi.1675398 · Cited by 1
Global Properties from Local Explanations with Concept Explanation Clusters
Haedecke, E., Akila, M., & von Rueden, L. (2025) · Communications in computer and information science · doi:10.1007/978-3-032-08317-3_1 · Cited by 1
TangledFeatures: Robust Feature Selection in Highly Correlated SpacesPreprint
Sunny, A. D. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2510.15005
“However, explanations often unstable in the presence of highly correlated structural features and may fail to yield biologically meaningful drivers [9].”
Takefuji, Y. (2025) · Journal of environmental chemical engineering · doi:10.1016/j.jece.2025.120094
Palar, P. S., Saves, P., Robani, M. D., Verstaevel, N., Garouani, M., Aligon, J., Shimoyama, K., Morlier, J., & Gaudou, B. (2026) · Archives of Computational Methods in Engineering · doi:10.1007/s11831-026-10600-z · Cited by 7
Lazzaroni, L. (2024) · CINECA IRIS Institutial Research Information System (University of Genoa) · doi:10.15167/lazzaroni-luca_phd2024-03-14 · Cited by 4
Salentin, F. G. (2023) · DuEPublico (University of Duisburg-Essen) · doi:10.17185/duepublico/78268
Κατανόηση των διαφορών μεταξύ ανθρώπινου και μηχανικά παραγόμενου κειμένου
Koukas, A. (2025) · DSpace - NTUA (National Technical University of Athens) · doi:10.26240/heal.ntua.28578
Best, C. (2023) · Enlighten: Theses (The University of Glasgow) · doi:10.5525/gla.thesis.83951
Wendel, K. (2021) · Research Repository (Delft University of Technology)
Computer audition for emotional wellbeing
Baird, A. (2021) · OPUS (Augsburg University)
Peixoto, D. T. D. S. (2024) · Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT)
Integrated Multi-Hazard and Vulnerability Modelling for Flood Risk Assessment in the US Gulf Coast
Liu, A. (2024) · UCL Discovery (University College London) · Cited by 3
Naaz, F. (2025) · University Library of Skövde (University of Skövde)
Foundations of Artificial Intelligence Frameworks: Notion and Limits of AGIPreprint
Bui, K. G. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2511.18517 · Cited by 1
Nduwayezu, G. (2025) · Lund University Publications (Lund University)
Isolating variable effects in supervised machine learning illustrated in educational data mining
Filho, R. L. C. S. (2024) · LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)
Head- and eye-based features for continuous core affect prediction
O'Dwyer, J. (. (2020) · Research@THEA
Explicabilidad de modelos predictivos para detección temprana de deterioro cognitivo con XAI
Sanz, A. C. (2025) · UNED repository
Torhan, S. A., Nasich, M. R. S., Sosa, M. A. C., & Kirchhoff, C. J. (2026) · Environmental Research Water · doi:10.1088/3033-4942/ae57d3
Takaki, K. (2022) · Institutional Repositories DataBase (IRDB)
Zhou, L., Zhou, C., Deng, J., Zhu, X., & Li, T. X. (2026) · CIN Computers Informatics Nursing · doi:10.1097/cin.0000000000001528
Parmentier, L. (2022) · theses.fr (ABES) · doi:10.70675/8bc9b46bz3ed6z4328z9537z2721cb116879 · Cited by 1
Prédiction de la défaillance de l’entreprise en appliquant des techniques de l’apprentissage machine
Nguyen, H. H. (2023) · theses.fr (ABES) · doi:10.70675/eab5e01fz3eb8z4faez94aezb1e782030559
Kim, M. (2026) · Research Square · doi:10.21203/rs.3.rs-9501010/v1
Feng, Y., & Park, J. (2026) · PeerJ Computer Science · doi:10.7717/peerj-cs.3868 · Cited by 1
Sohail, M., Yar, H., & Kim, H. S. (2026) · Mathematics · doi:10.3390/math14101666 · Cited by 1
A Step Towards Inherently Interpretable Causal Machine Learning Models For Decision SupportPreprint
Gonzalez, D. Z. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.24348
“…regardless of whether post-hoc methods or inherently interpretable models are used, their interpretations only reveal how ML models exploit correlations in data to maximize prediction performance, which should not be mistaken with causal insights (Molnar et al., 2022; Zschech et al., 2025).”
A Causal Argumentation Method for Explainability of Machine Learning ModelsPreprint
Salgado, H., Kendall, M., & Ceberio, M. C. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.21758
“While these approaches enhance transparency, they primarily capture correlations rather than causal or dialectical relationships among features [14].”
Sumiati, A. I., Hariguna, T., & Barkah, A. S. (2026) · International Conference on Current Research in Artificial Intelligence and Data Science (ICCRAIDS) · doi:10.1109/iccraids67816.2026.11519648
“Eigenvector centrality appears weaker and less consistently directional, matching the permutation indication that it contributes little incremental value once other centralities are present [8], [12].”
InfluentialPittala, R. B., Kiran, M. A., Sharma, N., Thaile, M., Singamaneni, K. K., & Byrapuneni, L. P. (2026) · Expert Systems with Applications · doi:10.1016/j.eswa.2026.133048
Pugliese, E., Lauzon, D., Dematteis, A., Rodani, S., Benedetti, G., & Gargini, A. (2026) · Hydrogeology Journal · doi:10.1007/s10040-026-03102-6
Fang, Q., Rahman, M., Wagener, T., & Pianosi, F. (2026) · doi:10.5194/egusphere-2026-3267
When Efficiency Shapes Fairness: Cost-Optimised Human-in-the-Loop Routing
Breene, V., O'Hagan, A., & Costello, F. (2026) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3805689.3812278
“However, a growing literature has documented fundamental limitations of post-hoc explanations, including instability, susceptibility to manipulation, and weak guarantees of faithfulness to underlying model behaviour [37, 48].”
When Explanations Deceive: Understanding Unintentional and Intentional Deception in XAI
Tekkesinoglu, S. (2026) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3805689.3812351
“Molnar et al. discuss general pitfalls of post-hoc model-agnostic methods, including feature dependencies, approximation uncertainties, and lack of causality [70].”
Wang, W., Ruan, H.-O., Butler, K. T., Kumada, A., & Sato, M. (2026) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.6c02120
Nakıp, M., Bulucu, P., Yildiz, O., & Guzelis, C. (2026) · IEEE Access · doi:10.1109/access.2026.3711446
“…techniques, such as SHAP (SHapley Additive exPlanations), LIME, or Grad-CAM have been adapted to visualize sensor channel saliency and temporal attention windows, these approaches often yield inconsistent explanations when applied to highly collinear multi-sensor datasets [134], [135], [136].”
Kang, J.-H., Jeon, Y.-J., So, J.-H., Moon, J., & Kim, S. (2026) · Scientific Reports · doi:10.1038/s41598-026-63302-8
“…methodological work on model-agnostic interpretation has cautioned that Shapley-value-based methods can lead to unjustified causal interpretations when confounding among predictors is not addressed 29 , a concern particularly relevant to observational data such as ours.”
Sharma, D. P. G. S. (2026) · International Journal of Emerging Technologies and Innovative Research · doi:10.48175/ijetir-6655
Im, H., Liu, Y., Thompson, J. C., Miranda-Contreras, G. I., Loke, K., Bacon, R., Lucas, L., Amador‐Noguez, D., Gray, S. M., & Venturelli, O. S. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.08.04.742850
Yao, Y., Cao, H., Xiao, Z., Liu, W., Feng, K., & Xie, X. (2026) · Journal of Cleaner Production · doi:10.1016/j.jclepro.2026.149238
Gerlings, J., Keramidis, P., Shollo, A., & Constantiou, I. (2026) · Information Systems Journal · doi:10.1111/isj.70059
Suresh, S., Vinod, B., Sudhakara, D., & Reddy, S. S. K. (2026) · Journal of Materials Engineering and Performance · doi:10.1007/s11665-026-14926-4
Electrodermal Activity as a Potential Diagnostic Biomarker for Alzheimer’s Disease: A Pilot Study
Pandian, S. S. R., Díaz, L. R. M., Ruiz-Triviño, J., Aguillon, D. F., & Posada-Quintero, H. F. (2026) · Sensors · doi:10.3390/s26175407
“For the same reason, the permutation importance estimates are unstable: model-agnostic importance measures are sensitive to sampling variability and to correlations among features, and are known to be unreliable at these sample sizes [58].”
Ahmad, A., & Busair, E.-K. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.08.25.747089
Shah, Q., Iqbal, F., Khan, K. A., Alabduljabbar, H., & Ahmad, F. (2026) · Journal of Building Pathology and Rehabilitation · doi:10.1007/s41024-026-00885-9
“…model behavior by identifying the most influential predictors, explaining their contribution to indi - vidual predictions, and illustrating both global and local response characteristics, thereby enhancing the reliability and transparency of the proposed hybrid machine learning framework [81 – 83].”
Eshetie, H. A., Alamirew, T., Beyene, A. A., & Sahlu, D. (2026) · Agricultural Water Management · doi:10.1016/j.agwat.2026.110747
Hoang, T.-H. T., Nguyen, H.-L., Hoang, H.-G. T., Bui, N. K., & Ly, H. (2026) · Buildings · doi:10.3390/buildings16173564
CAR-MIL: Counterfactual Attention Regularization for Multiple Instance Learning
Chraki, I., Marza, P., Christodoulidis, S., & Vakalopoulou, M. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37574-2_36 · Cited by 1
“Despite progress, many of these methods are constrained by computational cost, limited scalability to large bags [7,19,27] or limited explanatory capacity: they typically highlight instances that support the predicted class, but do not reveal evidence that contradicts or challenges the decision…”
Ahmed, N., Ahmed, R. M., & Teodoriu, C. (2026) · Materials Today Communications · doi:10.1016/j.mtcomm.2026.116122
Musto, H. (2026) · Goldsmiths (University of London) · doi:10.25602/gold.00040930
Beltiukov, R. (2026) · eScholarship (California Digital Library)
Wolsza, M., MacFadyen, S., Baade, J., van Niekerk, A., Coetsee, C., Singh, J., Davies, A. B., & Schmullius, C. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7520280
Kim, G., Han, Y., Kim, D., & Choi, H.-S. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7545930
Folded Hadamard A2D2E: Query-Efficient Main-Effect Estimation for High-Dimensional Black-Box ModelsPreprint
Zhao, Z. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7548820
Duyan, K. R., Duyan, J. M. M., Paloy, V. A. A., Paloy, E. A., Manuel, M. S., Aguilar, M. J., Lucena, J. R. G., Lasap, R. A., & Jesus, L. C. M. D. (2026) · International Journal of Artificial Intelligence and Machine Learning · doi:10.51483/ijaiml.6.12s.2026.436-446
Edeh, T., Olupayimo, E. O., & Bolarinwa, K. K. (2025) · Nigerian Journal of Business Education (NIGJBED)
Marengo, G. (n.d.) · Institutional Research Information System University of Turin (University of Turin)
Towards A Multi-Framework Approach To Explainable Artificial Intelligence
Salgado, H. (2026) · DigitalCommons@UTEP (The University of Texas at El Paso)
Metadata
- Authors
- Molnar, C., König, G., Herbinger, J., Freiesleben, T., Dandl, S., Scholbeck, C. A., Casalicchio, G., Grosse-Wentrup, M., & Bischl, B.
- Affiliations
- Leibniz Institute for Prevention Research and Epidemiology - BIPS (DE)
- University of Vienna (AT)
- Ludwig-Maximilians-Universität München (DE)
- Corresponding author
- Molnar, Christoph
- ORCID
- Venue
- Lecture Notes in Computer Science
- Publisher
- Springer International Publishing
- Published
- 2022
- Pages
- 39-68
- ISSN
- 0302-9743 (ISSN-L)
- 1611-3349
- OpenAlex
- W3194357095
- Cited by
- 211
- References
- 130
- Citations by year
- 2026: 41 · 2025: 56 · 2024: 48 · 2023: 51 · 2022: 11
- FWCI
- 22.95
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor, 106 d)
- Locations
- Lecture notes in computer science (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University) (cc-by)
- Funders
- Deutsche Forschungsgemeinschaft (437611051)
- Bundesministerium für Bildung und Forschung (01IS18036A)
- Graduate School of Systemic Neurosciences
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Physical Sciences › Computer Science › Software › Software Testing and Debugging Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
- Keywords
- model-agnostic interpretability, explainable machine learning, partial dependence plots, permutation importance, Shapley value, feature interaction, uncertainty estimation, high-dimensional data, causal interpretation
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.