Accurate intelligible models with pairwise interactions
The DOI matches its Crossref record; 21 of the 22 references it cites were found in a registry or an index.
Abstract
Standard generalized additive models (GAMs) usually model the dependent variable as a sum of univariate models. Although previous studies have shown that standard GAMs can be interpreted by users, their accuracy is significantly less than more complex models that permit interactions.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
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?
In the citing paper
“It is designed to have high accuracy compared to the state-of-the-art machine learning models while keeping the intelligibility, and explainability [13].”
From An Empirical Comparison of Interpretable Models to Post-Hoc Explanations (2023) · cites this paper as [13]
Citations
Cites
Greedy function approximation: A gradient boosting machine
Friedman, J. H. (2001) · The Annals of Statistics · doi:10.1214/aos/1013203451 · Cited by 30,731
Generalized Additive Models: An Introduction with R
Wood, S. N. (2017) · doi:10.1201/9781315370279 · Cited by 13,852
Introduction to information retrieval
(2009) · Choice Reviews Online · doi:10.5860/choice.46-2715 · Cited by 12,278
Brown, R. A., Hastie, T. J., & Tibshirani, R. J. (1991) · Biometrics · doi:10.2307/2532174 · Cited by 8,176
An introduction to variable and feature selection
GuyonIsabelle, & ElisseeffAndré (2003) · Journal of Machine Learning Research · doi:10.5555/944919.944968 · Cited by 7,715
Koyak, R. A., Hastie, T. J., & Tibshirani, R. J. (1991) · Journal of the American Statistical Association · doi:10.2307/2290538 · Cited by 7,596
Wood, S. N. (2006) · doi:10.1201/9781420010404 · Cited by 4,953
Hastie, T., & Tibshirani, R. (2014) · Wiley StatsRef: Statistics Reference Online · doi:10.1002/9781118445112.stat03141 · Cited by 3,156
An Empirical Comparison of Voting Classification Algorithms: Bagging, Boosting, and Variants
Bauer, E., & Kohavi, R. (1999) · Machine Learning · doi:10.1023/a:1007515423169 · Cited by 2,638
Wood, S. N. (2003) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/1467-9868.00374 · Cited by 2,445
Research works printed but not linked: 2
- 10.5555/1394399 · doi:10.5555/1394399
- 10.5555/1622620.1622633 · doi:10.5555/1622620.1622633
Cited by
Items 101–200 of 565
Accurate, interpretable predictions of materials properties within transformer language models
Korolev, V., & Protsenko, P. V. (2023) · Patterns · doi:10.1016/j.patter.2023.100803 · Cited by 49
“Linear regression models and their extension, generalized additive models[32, 33], provide weight coefficients as an importance metric of relevant features[34, 35, 36].”
BackgroundPachauri, N. (2024) · Fuel · doi:10.1016/j.fuel.2024.131421 · Cited by 42
Computational Predictions of Nonclinical Pharmacokinetics at the Drug Design Stage
Stoyanova, R., Katzberger, P., Komissarov, L., Khadhraoui, A., Sach-Peltason, L., Zbinden, K. G., Schindler, T., & Manevski, N. (2023) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.2c01134 · Cited by 41
Jiang, C., Yan, B., Goldsmith, B. R., & Linic, S. (2025) · Nature Catalysis · doi:10.1038/s41929-025-01417-3 · Cited by 50
“…such as Random Forest and Boosted Trees, while being highly intelligible and explainable. iGAM is a fast implementation of the GA2M algorithm 30 , written in C++ and Python, with the following equation (5): Here, g is a link function that adapts the GAM to different settings, such as…”
InfluentialBig Data Visualization and Analytics: Future Research Challenges and Emerging Applications
Andrienko, G., Andrienko, N. V., Drucker, S. M., Fekete, J., Fisher, D., Idreos, S., Kraska, T., Li, G., Ma, K., Mackinlay, J. D., et al. (2020) · HAL · Cited by 38
Locally interpretable tree boosting: An application to house price prediction
Hjort, A., Scheel, I., Sommervoll, D. E., & Pensar, J. (2023) · Decision Support Systems · doi:10.1016/j.dss.2023.114106 · Cited by 44
“Although most humans find it easier to visually analyze a one-dimensional shape function f j ( x j ) than a heat map f kl ( x k , x l ), Lou et al. (2013) argues that this is still intelligible.”
BackgroundMethodInfluentialTowards trustworthy AI in Industry 5.0: Ante-hoc interpretability with deep learning
Zhou, L., Chen, L., Zhu, T., Su, X., Dong, Y., Wen, L., Liu, H., Lu, X., Leng, J., Liu, Q., et al. (2026) · Computers & Industrial Engineering · doi:10.1016/j.cie.2026.111805 · Cited by 55
Axiomatic Interpretability for Multiclass Additive Models
Zhang, X., Tan, S., Koch, P., Lou, Y., Chajewska, U., & Caruana, R. (2019) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3292500.3330898 · Cited by 50
“Another line of research aims at building interpretable models from the ground-up, such as rule lists [18, 30], scoring systems [31], decision sets [17], and additive models [21].”
BackgroundSelf-citationOptimization of Thermal Conductivity of NanoPCM-Based Graphene by Response Surface Methodology
Aslfattahi, N., Zendehboudi, A., Saidur, R., Sabri, M. F. M., Said, S. M., Arifutzzaman, A., & Sidik, N. A. C. (2020) · Journal of Advanced Research in Fluid Mechanics and Thermal Sciences · doi:10.37934/arfmts.75.3.108125 · Cited by 38
“The other one was pre-specifying the all interaction forms of interest then using lasso to simultaneously select the important points [36].”
MethodTsang, M., Liu, H., Purushotham, S., Murali, P., & Liu, Y. (2018) · neural information processing systems · Cited by 70
Gill, N., Hall, P., Montgomery, K., & Schmidt, N. (2020) · Information · doi:10.3390/info11030137 · Cited by 36
“XNNs also bear a strong resemblance to generalized additive models (GAMs) and so-called explainable boosting machines (EBMs or GA2Ms), which consider main effects and a small number of two-way interactions and may also incorporate boosting into their training [14,29].”
BackgroundMethodMaking Neural Networks Interpretable with Attribution: Application to Implicit Signals Prediction
Afchar, D., & Hennequin, R. (2020) · ACM Conference on Recommender Systems (RecSys) · doi:10.1145/3383313.3412253 · Cited by 38
“This formulation is closely related to the mixture of experts [23, 24], allowing us to train our model jointly instead of iteratively as in [29].”
BackgroundMethodInfluentialExplainable AI and Adoption of Financial Algorithmic Advisors: An Experimental Study
David, D. B., Resheff, Y. S., & Tron, T. (2021) · Proceedings of the AAAI/ACM Conference on AI Ethics and Society · doi:10.1145/3461702.3462565 · Cited by 79
“learning, to develop fast and accurate techniques which explain black-box or opaque models [15, 20, 22], or train AI which is more interpretable by nature [1, 19].”
BackgroundSarica, A., Quattrone, A., & Quattrone, A. (2022) · Brain Imaging and Behavior · doi:10.1007/s11682-022-00688-9 · Cited by 45
“…is based on standard Generalized Additive Models (GAMs) (Hastie & Tibshi-rani, 1990), which accuracy is improved by adding pairwise interactions (Lou et al., 2013), taking the name of GA 2 Ms and the form: where x i = ( x i 1 , … , x ip ) is the feature vector with p features, y i the response,…”
BackgroundMethodInfluentialNew Findings From Explainable SYM‐H Forecasting Using Gradient Boosting Machines
Iong, D., Chen, Y., Tóth, G., Zou, S., Pulkkinen, T. I., Ren, J., Camporeale, E., & Gombosi, T. I. (2022) · Space Weather · doi:10.1029/2021sw002928 · Cited by 47
“Recent efforts have been made to close this gap, starting with additive models that incorporate two-way feature interactions (Lou et al., 2013).”
BackgroundBoxer: Interactive Comparison of Classifier Results
Gleicher, M. L., Barve, A., Yu, X., & Heimerl, F. (2020) · Computer Graphics Forum · doi:10.1111/cgf.13972 · Cited by 47
On parameter interpretability of phenomenological-based semiphysical models in biology
Lema-Pérez, L., Muñoz‐Tamayo, R., Garcia‐Tirado, J., & Álvarez, H. (2019) · Informatics in Medicine Unlocked · doi:10.1016/j.imu.2019.02.002 · Cited by 43
Interpretable Ranking with Generalized Additive Models
Zhuang, H., Wang, X., Bendersky, M., Grushetsky, A., Wu, Y., Mitrichev, P., Sterling, E., Bell, N., Ravina, W., & Qian, H. (2021) · ACM International Conference on Web Search and Data Mining (WSDM) · doi:10.1145/3437963.3441796 · Cited by 59
“However, we are aware that there are numerous directions to further develop and understand ranking GAMs: (1) Incorporating limited pairwise feature interactions to improve model accuracy with reasonable interpretability (similar to [32]); (2) Enabling users to dictate constraints such as monotonicity on sub-models; (3…”
BackgroundZhang, W., Deng, Z., Zhang, T., Choi, K., Wang, J., & Wang, S. (2021) · IEEE Transactions on Fuzzy Systems · doi:10.1109/tfuzz.2021.3099696 · Cited by 35
Considerations when learning additive explanations for black-box models
Tan, S., Hooker, G., Koch, P., Gordo, A., & Caruana, R. (2023) · Machine Learning · doi:10.1007/s10994-023-06335-8 · Cited by 176
“Pairwise interactions can also be modeled explicitly [44], in the form of ℎ 𝑖𝑖 ′ ( 𝑥 ′ ) added to Equation 7.”
Self-citationBDT: Gradient Boosted Decision Tables for High Accuracy and Scoring Efficiency
Lou, Y., & Obukhov, M. (2017) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3097983.3098175 · Cited by 45
“have developed a fast method of detecting pairwise feature interactions for intelligibility using 2-dimensional decision tables, although not explicitly stated [15].”
MethodSelf-citationExplainable Boosting Machine for Predicting Alzheimer’s Disease from MRI Hippocampal Subfields
Sarica, A., Quattrone, A., & Quattrone, A. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-86993-9_31 · Cited by 33
Casimiro‐Soriguer, C. S., Loucera, C., Peña-Chilet, M., & Dopazo, J. (2022) · Scientific Reports · doi:10.1038/s41598-021-04182-y · Cited by 44
A Roadmap of Explainable Artificial Intelligence: Explain to Whom, When, What and How?
Wang, Z., Huang, C., & Yao, X. (2024) · ACM Transactions on Autonomous and Adaptive Systems · doi:10.1145/3702004 · Cited by 52
“Global FAE describes the overall significance of each feature used by the ML model [120], with existing methods including [84, 88, 128, 129, 209, 220].”
Distill-and-Compare: Auditing Black-Box Models Using Transparent Model Distillation
Tan, S., Caruana, R., Hooker, G., & Lou, Y. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1710.06169 · Cited by 22
Self-citationZilker, S., Weinzierl, S., Kraus, M., Zschech, P., & Matzner, M. (2024) · Health Care Management Science · doi:10.1007/s10729-024-09673-8 · Cited by 29
“Traditionally, shape functions are only computed for static features [e.g., 13, 16, 51–55].”
BackgroundInfluentialKaratekin, T., Sancak, S., Çelik, G., Topçuoğlu, S., Karatekın, G., Kirci, P., & Okatan, A. (2019) · International Conference on Deep Learning and Machine Learning in Emerging Applications (Deep-ML) · doi:10.1109/deep-ml.2019.00020 · Cited by 36
“[2,5] Thus, as we make our model more complex, we will add the pairwise interactions one by one.”
MethodExplainable Artificial Intelligence for COVID-19 Diagnosis Through Blood Test Variables
Thimoteo, L. M., Vellasco, M. M. B. R., Amaral, J. L. M., Figueiredo, K., Yokoyama, C. L., & Marques, E. (2022) · Journal of Control Automation and Electrical Systems · doi:10.1007/s40313-021-00858-y · Cited by 39
“EBM is a fast implementation of the Lou et al. (2013).”
Formal Hypothesis Tests for Additive Structure in Random Forests
Mentch, L., & Hooker, G. (2016) · Journal of Computational and Graphical Statistics · doi:10.1080/10618600.2016.1256817 · Cited by 38
“…including a method based on marginal integration by Linton (1995), a wavelet method suggested by Amato and Antoniadis (2001), a tree-based method by Lou et al. (2013), and the most popular class based on backfit-ting algorithms as found in Buja, Hastie, and Tibshirani (1989), Opsomer and Ruppert…”
MethodSelf-citationDetecting Bias in Black-Box Models Using Transparent Model Distillation.Preprint
Tan, S., Caruana, R., Hooker, G., & Lou, Y. (2017) · arXiv (Cornell University) · Cited by 36
Self-citationInterpretability and Transparency in Artificial Intelligence
Mittelstadt, B. (2022) · Oxford University Press eBooks · doi:10.1093/oxfordhb/9780198857815.013.20 · Cited by 41
Mariotti, E., Alonso, J. M., & Gatt, A. (2023) · Information Fusion · doi:10.1016/j.inffus.2023.101882 · Cited by 33
Pisu, F., Williamson, B. J., Nardi, V., Paraskevas, K. I., Puig, J., Vagal, A. S., Rubeis, G. D., Porcu, M., Cau, R., Benson, J. C., et al. (2024) · Circulation Cardiovascular Imaging · doi:10.1161/circimaging.123.016274 · Cited by 27
“23 The GB-GAM has been effectively used in other studies with cardiac data.”
MethodPineda-Jaramillo, J., Munoz, C., Mesa-Arango, R., Gonzalez-Calderon, C., & Lange, A. (2024) · Research in Transportation Business & Management · doi:10.1016/j.rtbm.2024.101161 · Cited by 43
An Interpretable Machine Learning Model for Predicting 10-Year Total Hip Arthroplasty Risk
Jang, S. J., Fontana, M. A., Kunze, K. N., Anderson, C. G., Sculco, T. P., Mayman, D. J., Jerabek, S. A., Vigdorchik, J. M., & Sculco, P. K. (2023) · The Journal of Arthroplasty · doi:10.1016/j.arth.2023.03.087 · Cited by 30
Ma, Y., Zhang, P., Duan, S., & Zhang, T. (2023) · Finance research letters · doi:10.1016/j.frl.2023.104305 · Cited by 31
Proposed Guidelines for the Responsible Use of Explainable Machine Learning
Hall, P., Gill, N., & Schmidt, N. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.03533 · Cited by 15
Shapley Explanation NetworksPreprint
Wang, R., Wang, X., & Inouye, D. I. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.02297 · Cited by 50
“In particular, GAM [25] and its extension [26, 27] construct models that can be displayed as series of line graphs or heatmaps.”
BackgroundA United States Fair Lending Perspective on Machine Learning
Hall, P., Cox, B., Dickerson, S., Kannan, A. R., Kulkarni, R., & Schmidt, N. (2021) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2021.695301 · Cited by 37
“, explainable boosting machines (EBMs, also known as GA2Ms) (Lou et al., 2013)]; constrained tree-based models [e.”
BackgroundMethodBarbado, A., & Corcho, Ó. (2022) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2022.105222 · Cited by 39
Liu, X., Shi, X.-Q., Peng, Z., & He, H.-D. (2023) · Sustainable Cities and Society · doi:10.1016/j.scs.2023.104524 · Cited by 29
Interpretable machine learning for creditor recovery rates
Nazemi, A., & Fabozzi, F. J. (2024) · Journal of Banking & Finance · doi:10.1016/j.jbankfin.2024.107187 · Cited by 24
Zeitlin, J. H., Parides, M. K., Lane, J. M., Russell, L., & Kunze, K. N. (2023) · Archives of Osteoporosis · doi:10.1007/s11657-023-01292-0 · Cited by 25
“To maximize validity, the GA2M also considered six pairwise interaction effects among the 14 risk factors [41].”
Explainable Fault Diagnosis of Oil-Immersed Transformers: A Glass-Box Model
Liao, W., Zhang, Y., Cao, D., Ishizaki, T., Yang, Z., & Yang, D. (2024) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2024.3350131 · Cited by 23
“As in previous publications [8], interpretability is defined as the ability to understand the contribution of individual features in a model.”
BackgroundMachine-learned patterns suggest that diversification drives economic development
Brummitt, C. D., Gómez-Liévano, A., Hausmann, R., & Bonds, M. H. (2020) · Journal of The Royal Society Interface · doi:10.1098/rsif.2019.0283 · Cited by 31
“ed by machines less tied to disciplinary paradigms. Rapidly advancing ways for machines to learn interpretable models bode well for followup studies. Pairwise interactions could be modeled using GA2M [42], or high-dimensional data could be tted with GAMs that both smooth the data and select terms, such as GAMSEL [18]…”
MethodMining Cross Features for Financial Credit Risk Assessment
Liu, Q., Liu, Z., Zhang, H., Chen, Y., & Zhu, J. (2021) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3459637.3482371 · Cited by 30
“ForLR,GA M [33] and LR models used in GBDT+LR [17], Au-toCross range of { 0 .”
MethodUsing explainable machine learning and fitbit data to investigate predictors of adolescent obesity
Kiss, O., Baker, F. C., Pálovics, R., Dooley, E. E., Gabriel, K. P., & Nagata, J. M. (2024) · Scientific Reports · doi:10.1038/s41598-024-60811-2 · Cited by 34
Pachauri, N., & Ahn, C. W. (2022) · IEEE Access · doi:10.1109/access.2022.3153720 · Cited by 27
“j ψjfj(Xj) (5) where, fj(i) = jth basis function 9j = parameter of jth basis function In order to improve the accuracy of the conventional GAM, a pairwise interactions are added to it, then equation (5) can be modified as [25].”
BackgroundTSFuse: automated feature construction for multiple time series data
Brabandere, A. D., Beéck, T. O. D., Hendrickx, K., Meert, W., & Davis, J. J. (2022) · Machine Learning · doi:10.1007/s10994-021-06096-2 · Cited by 27
Interpretability, Then What? Editing Machine Learning Models to Reflect Human Knowledge and Values
Wang, Z. J., Kale, A. M., Nori, H., Stella, P., Nunnally, M., Chau, D. H., Vorvoreanu, M., Vaughan, J. W., & Caruana, R. (2022) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3534678.3539074 · Cited by 44
“2): the first interactive system to empower domain experts and data scientists to easily and responsibly edit the weights of generalized additive models (GAMs) [12, 18, 19], a state-of-the-art interpretable model [26].”
BackgroundSelf-citationHegselmann, S., Ertmer, C., Volkert, T., Gottschalk, A., Dugas, M., & Varghese, J. (2022) · Frontiers in Medicine · doi:10.3389/fmed.2022.960296 · Cited by 32
“(34) and use the logit link function for dichotomous classifications analogous to LR (see equation 2, note that the logit function was moved to the right side).”
MethodResultInvestigating Searchers’ Mental Models to Inform Search Explanations
Thomas, P., Billerbeck, B., Craswell, N., & White, R. W. (2019) · ACM Transactions on Information Systems · doi:10.1145/3371390 · Cited by 25
“There have been many attempts to develop interpretable and intelligible ML models to tackle specific challenges [Caruana et al. 2015; Jung et al. 2017; Lakkaraju et al. 2017; Letham et al. 2015; Lou et al. 2013].”
BackgroundCalibrate: Interactive Analysis of Probabilistic Model Output
Xenopoulos, P., Rulff, J., Nonato, L. G., Barr, B., & Silva, C. T. (2022) · IEEE Transactions on Visualization and Computer Graphics · doi:10.1109/tvcg.2022.3209489 · Cited by 25
“For our work, we use Explainable Boosting Machines (EBMs) [29, 30, 36], which are available in the interpret Python package.”
MethodSarica, A., Quattrone, A., & Quattrone, A. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-15037-1_11 · Cited by 20
GAM Coach: Towards Interactive and User-centered Algorithmic Recourse
Wang, Z. J., Vaughan, J. W., Caruana, R., & Chau, D. H. (2023) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3544548.3580816 · Cited by 35
“, �} or �� � (�� , � � ) for pairwise interactions between features [52] is learned using gradient-boosted trees [51].”
MethodSelf-citationCaleca, F., Confuorto, P., Raspini, F., Segoni, S., Tofani, V., Casagli, N., & Moretti, S. (2024) · The Science of The Total Environment · doi:10.1016/j.scitotenv.2024.175277 · Cited by 22
Moving from ‘black box’ to ‘glass box’ Artificial Intelligence in Manufacturing with XMANAI
Lampathaki, F., Agostinho, C., Glikman, Y., & Sesana, M. (2021) · IEEE International Conference on Engineering, Technology and Innovation (ICE/ITMC) · doi:10.1109/ice/itmc52061.2021.9570236 · Cited by 28
Industrial AI in condition-based maintenance: A case study in wooden piece manufacturing
Martí-Puig, P., Touhami, I. A., Perarnau, R. C., & Serra-Serra, M. (2024) · Computers & Industrial Engineering · doi:10.1016/j.cie.2024.109907 · Cited by 34
Ethical Implementation of Artificial Intelligence to Select Embryos in In Vitro Fertilization
Afnan, M. A. M., Rudin, C. D., Conitzer, V., Savulescu, J., Mishra, A., Liu, Y., & Afnan, M. (2021) · Proceedings of the AAAI/ACM Conference on AI Ethics and Society · doi:10.1145/3461702.3462589 · Cited by 42
“For tabular data, neural networks and other forms of black-box models do not seem to provide additional accuracy, which means that optimized medical scoring systems might be as accurate as one could get (depending on the dataset) [66, 67].”
BackgroundMicromodels for Efficient, Explainable, and Reusable Systems: A Case Study on Mental Health
Lee, A. S., Kummerfeld, J. K., An, L. C., & Mihalcea, R. F. (2021) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2021.findings-emnlp.360 · Cited by 31
StrategyAtlas: Strategy Analysis for Machine Learning Interpretability
Collaris, D., & van Wijk, J. J. (2022) · IEEE Transactions on Visualization and Computer Graphics · doi:10.1109/tvcg.2022.3146806 · Cited by 21
“There are two main approaches in the machine learning community to produce insights: either creating inherently interpretable models (e.g., GAM [5] and CORELS [6]), or explaining models post-hoc, using an exter-nal method (e.g., LIME [3] and SHAP [4]).”
Koca, M. B., Nouranı, E., ABBASOĞLU, F., Karadenız, İ., & Sevilgen, F. E. (2022) · Computational Biology and Chemistry · doi:10.1016/j.compbiolchem.2022.107755 · Cited by 20
Interpretability and Explainability in Machine Learning
Tsang, W. K., & Benoit, D. F. (2022) · Intelligent systems reference library · doi:10.1007/978-3-031-11593-6_5 · Cited by 25
Priya Dharshini, S., Ramkumar, K., Venkatesh, S., Narasimhan, K., & Adalarasu, K. (2023) · International Conference on Advanced Computing and Communication Systems (ICACCS) · doi:10.1109/icaccs57279.2023.10113001 · Cited by 22
Yao, L., Yang, Z., Zhang, Z., Tang, S., Shen, B., & Zeng, J. (2023) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2023.3323006 · Cited by 22
“To reduce the computational complexity, a tree-based model is built to filter the useless interactions considering the heredity constraint [31].”
MethodCai, J., Hu, W., Ma, J., Si, A., Chen, S., Gong, L., Zhang, Y., Yan, H., Chen, F., & Initiative, A. D. N. (2023) · Brain Sciences · doi:10.3390/brainsci13111535 · Cited by 24
“The EBM algorithm extends GAMs by adding pairwise interactions between relevant variables, taking the name of GA 2 M, which can be defined as in Equation (2), where the s i j x i , x j donates the pairwise interactions [23].”
MethodPower to the People! - Meta-Algorithmic Modelling in Applied Data Science
Spruit, M., & Jagesar, R. (2016) · International Conference on Knowledge Discovery and Information Retrieval · doi:10.5220/0006081604000406 · Cited by 16
“A variant of linear modelling is applied known as generalized additive modelling (GAM) enriched with information on pairwise interactions between features (Lou et al., 2013).”
MethodSelf-regulated Learning in the Digitally Enhanced Science Classroom: Toward an Early Warning System
Kubsch, M., Strauß, S., Grimm, A., Gombert, S., Drachsler, H., Neumann, K., & Rummel, N. (2025) · Educational Psychology Review · doi:10.1007/s10648-025-10011-9 · Cited by 27
“In this case, more explainable models (e.g., explainable boosting machines, Lou et al., 2013) may shed light on the question which degrees of the different emotions are particularly conducive to learning.”
Arithmetic Feature Interaction Is Necessary for Deep Tabular Learning
Cheng, Y., Hu, R., Ying, H., Shi, X., Wu, J., & Lin, W. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i10.29033 · Cited by 29
Instance-Level Explanations for Fraud Detection: A Case StudyPreprint
Collaris, D., Vink, L. M., & van Wijk, J. J. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1806.07129 · Cited by 32
“Sensitivity analysis Another approach to analyze features is through sensitivity analysis (Cortez & Embrechts, 2013; Goldstein et al., 2015; Friedman, 2001; Welling et al., 2016; Krause et al., 2016; Lou et al., 2013).”
NODE-GAM: Neural Generalized Additive Model for Interpretable Deep LearningPreprint
Chang, C., Caruana, R., & Goldenberg, A. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.01613 · Cited by 86
“NODE outperforms all GAMs and GA2Ms substantially on Higgs and Year, suggesting both datasets might have important higher-order feature interactions.”
BackgroundMethodInfluentialSelf-citationJović, A., Frid, N., Brkić, K., & Cifrek, M. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2025.108153 · Cited by 27
Jayasundara, S., Indika, A., & Herath, D. (2022) · International Conference on Advanced Research in Computing (ICARC) · doi:10.1109/icarc54489.2022.9753867 · Cited by 18
Xie, W., Kimura, M., Takaki, K., Asada, Y., Iida, T., & Jia, X. (2022) · Water Resources Research · doi:10.1029/2021wr030493 · Cited by 22
“These include sensitivity analysis (Cortez & Embrechts, 2013), generalized additive (Caruana et al., 2015; Lou et al., 2013), local interpretable model-agnostic explanatory (Ribeiro et al., 2016), layer-wise relevance propagation (Montavon et al., 2018; Sturm et al., 2016), and other models.”
MethodInterpretable Machine Learning Models for Phase Prediction in Polymerization-Induced Self-Assembly
Lu, Y., Yalçın, D., Pigram, P. J., Blackman, L. D., & Boley, M. (2023) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.3c00460 · Cited by 18
Vasilaras, A., Papadoudis, N., & Rizomiliotis, P. (2024) · Forensic Science International Digital Investigation · doi:10.1016/j.fsidi.2024.301737 · Cited by 35
Integrating Fuzzy C-Means Clustering and Explainable AI for Robust Galaxy Classification
Díaz, G. M., Medina, R. G., & Jiménez, J. A. A. (2024) · Mathematics · doi:10.3390/math12182797 · Cited by 19
“Model-dependent Explainability [34] techniques are specific to certain types of models and leverage their internal structure to provide explanations.”
Moore, A., Murdock, V., Cai, Y., & Jones, K. (2018) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3209978.3210151 · Cited by 17
“[2, 6, 7] GAMs are especially suited for model explanations because they lend themselves to visualizing the distributions of the values of the feature functions.”
BackgroundMorrison, C. P., Huckvale, K., Corish, B., Banks, R., Grayson, M., Dorn, J. F., Sellen, A. J., & Lindley, S. (2018) · ACM Transactions on Interactive Intelligent Systems · doi:10.1145/3181670 · Cited by 18
“Some research is closing this gap, but still unable to deal with large-dimensional data [32].”
BackgroundDecision tree boosted varying coefficient models
Zhou, Y., & Hooker, G. (2022) · Data Mining and Knowledge Discovery · doi:10.1007/s10618-022-00863-y · Cited by 19
“An example is given Lou et al. (2012, 2013) where trees were used for univariate functions of each feature in a GAMmodel; later used in Tan et al. (2018) for model distillation.”
MethodSelf-citationQuality aspects of annotated data: A research synthesis
Beck, J. (2023) · AStA Wirtschafts- und Sozialstatistisches Archiv · doi:10.1007/s11943-023-00332-y · Cited by 29
“While some studies observe improved data quality through performance-based payments (Ho et al. 2015), others could not confirm the existence of this relationship (Shaw et al. 2011; Lou et al. 2013).”
From bias to black boxes: understanding and managing the risks of AI – an actuarial perspective
du Preez, V., Bennet, S., Byrne, M., Couloumy, A., Das, A., Dessain, J., Galbraith, R., King, P., Mutanga, V., Schiller, F., et al. (2024) · British Actuarial Journal · doi:10.1017/s1357321724000060 · Cited by 15
Transparent Model Distillation.Preprint
Tan, S., Caruana, R., Hooker, G., & Gordo, A. (2018) · arXiv (Cornell University) · Cited by 15
Self-citationLearning outside the Black-Box: The pursuit of interpretable modelsPreprint
Crabbé, J., Zhang, Y., Zame, W. R., & van der Schaar, M. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.08596 · Cited by 39
“ntary material. 5 Related work Table 1: Symbolic Pursuit and the state-of-the-art interpretability methods. Algorithm Feature Importance Feature Interaction Model Independent Global Parsimonious GA2M [30] LIME [37] SHAP [32] DeepLIFT [40] L2X [7] NIT [42] INVASE [45] Symb. Metamodel [1] Symbolic Pursuit In this sectio…”
ResultDevelopment of a Novel Transformation of Spiking Neural Classifier to an Interpretable Classifier
Jeyasothy, A., Sundaram, S., Ramasamy, S., & Sundararajan, N. (2022) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2022.3181181 · Cited by 15
“Commonly used approaches to fit the shape functions in GAMs are by using linear functions [16], [17], using another base learner such as decision trees [13], [18], [19] or using neural networks [20].”
MethodTrustworthy Academic Risk Prediction with Explainable Boosting Machines
Dsilva, V., Schleiß, J., & Stober, S. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-36272-9_38 · Cited by 20
Interpretable Predictive Models to Understand Risk Factors for Maternal and Fetal Outcomes
Bosschieter, T. M., Xu, Z., Lan, H., Lengerich, B. J., Nori, H., Painter, I. S., Souter, V., & Caruana, R. (2023) · Journal of Healthcare Informatics Research · doi:10.1007/s41666-023-00151-4 · Cited by 28
“Additionally, EBMs can be trained to automatically detect important pairwise interaction terms to further boost accuracy while preserving intelligibility [31].”
Self-citationExplainable generalized additive neural networks with independent neural network training
Ortega-Fernandez, I., Sestelo, M., & Villanueva, N. M. (2023) · Statistics and Computing · doi:10.1007/s11222-023-10320-5 · Cited by 19
“Chang et al. (2021) introduced NODE-GAM, a GAM based on a deep learning version of tree-based GAMs and GA 2 M (Lou et al. 2013).”
MethodLeveraging interpretable machine learning in intensive care
Bohlen, L., Rosenberger, J., Zschech, P., & Kraus, M. (2024) · Annals of Operations Research · doi:10.1007/s10479-024-06226-8 · Cited by 24
“In more recent approaches bagged and boosted tree ensembles (Lou et al., 2012, 2013) are used.”
AI-driven valuation: a new era for real estate appraisal
Topraklı, A. Y. (2024) · Journal of European real estate research · doi:10.1108/jerer-04-2024-0031 · Cited by 37
Solving Non-image Learning Problems by Mapping to Images
Kovalerchuk, B., Agarwal, B., & Kall, D. C. (2020) · International Conference on Information Visualisation (IV) · doi:10.1109/iv51561.2020.00050 · Cited by 13
“While it requires more experimenting with CPC-R and CNN architectures, extensive relevant experiments with positive results already have been conducted for another pairs-based method [7].”
BackgroundMethodDeepCM: Deep neural networks to improve accuracy prediction of database cost models
Ouared, A., Chadli, A., & Daoud, M. A. (2021) · Concurrency and Computation Practice and Experience · doi:10.1002/cpe.6724 · Cited by 14
Explicabilidad (de la inteligencia artificial)
Alcarazo, L. O. D. Z. (2022) · EUNOMÍA Revista en Cultura de la Legalidad · doi:10.20318/eunomia.2022.6819 · Cited by 39
Binary ECG Classification Using Explainable Boosting Machines for IoT Edge Devices
Xiaolin, L., Qingyuan, W., Panicker, R. C., Cardiff, B., & John, D. (2022) · IEEE International Conference on Electronics, Circuits and Systems (ICECS) · doi:10.1109/icecs202256217.2022.9970834 · Cited by 14
“Explainable Boosting Machine (EBM) is a tree-based, cyclic gradient boosting Generalized Additive Model (GAM) with automatic interaction detection [12].”
BackgroundInfluentialCounterNet: End-to-End Training of Prediction Aware Counterfactual Explanations
Guo, H., Nguyen, T. H., & Yadav, A. (2023) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3580305.3599290 · Cited by 30
“) that are intrinsically interpretable [6, 22, 23, 38]; and (ii) applying “black-box" ML models, and explaining their predictions in a post-hoc manner [7, 35, 53].”
MethodAuditing Black-Box Models Using Transparent Model Distillation With Side InformationPreprint
Tan, S., Caruana, R., Hooker, G., & Lou, Y. (2017) · arXiv (Cornell University) · Cited by 14
Self-citationFrom point to probabilistic gradient boosting for claim frequency and severity prediction
Chevalier, D., & Côté, M. (2025) · European Actuarial Journal · doi:10.1007/s13385-025-00428-5 · Cited by 20
“EGBM follows a two-stage construction approach [14], learning the main effects first and the interaction effects second.”
InfluentialInterpretable machine learning algorithms to predict leaf senescence date of deciduous trees
Gao, C., Wang, H., & Ge, Q. (2023) · Agricultural and Forest Meteorology · doi:10.1016/j.agrformet.2023.109623 · Cited by 15
Ngene, G. M., & Wang, J. (2023) · Accounting and Finance · doi:10.1111/acfi.13154 · Cited by 13
Liu, X., Shi, X.-Q., Li, X., & Peng, Z. (2022) · Journal of Cleaner Production · doi:10.1016/j.jclepro.2022.134494 · Cited by 12
Metadata
- Authors
- Lou, Y., Caruana, R., Gehrke, J., & Hooker, G.
- Affiliations
- Cornell University (US)
- Microsoft (United States) (US)
- Venue
- Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining
- Publisher
- ACM
- Published
- 2013-08-11
- Pages
- 623-631
- OpenAlex
- W2046945713
- Cited by
- 564
- References
- 22
- Citations by year
- 2026: 66 · 2025: 78 · 2024: 88 · 2023: 87 · 2022: 65 · 2021: 57 · 2020: 45 · 2019: 29 · 2018: 24 · 2017: 11 · 2016: 8 · 2015: 4 · 2014: 2
- FWCI
- 3.01
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Applications
- Physical Sciences › Computer Science › Computer Graphics and Computer-Aided Design › Data Visualization and Analytics
- Keywords
- generalized additive models, pairwise interactions, model interpretability, model accuracy
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.