{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,14]],"date-time":"2026-04-14T16:10:51Z","timestamp":1776183051144,"version":"3.50.1"},"reference-count":90,"publisher":"Springer Science and Business Media LLC","issue":"3","license":[{"start":{"date-parts":[[2025,3,17]],"date-time":"2025-03-17T00:00:00Z","timestamp":1742169600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2025,3,17]],"date-time":"2025-03-17T00:00:00Z","timestamp":1742169600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"DOI":"10.13039\/501100001659","name":"Deutsche Forschungsgemeinschaft","doi-asserted-by":"publisher","award":["EXC 2046\/1 project-ID: 390685689"],"award-info":[{"award-number":["EXC 2046\/1 project-ID: 390685689"]}],"id":[{"id":"10.13039\/501100001659","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001659","name":"Deutsche Forschungsgemeinschaft","doi-asserted-by":"publisher","award":["KI-FOR 5363, project ID: 459422098"],"award-info":[{"award-number":["KI-FOR 5363, project ID: 459422098"]}],"id":[{"id":"10.13039\/501100001659","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100002347","name":"Bundesministerium f\u00fcr Bildung und Forschung","doi-asserted-by":"publisher","award":["01IS14013A-E, 01GQ1115, 01GQ0850, 01IS18056A, 01IS18025A, 01IS18037A"],"award-info":[{"award-number":["01IS14013A-E, 01GQ1115, 01GQ0850, 01IS18056A, 01IS18025A, 01IS18037A"]}],"id":[{"id":"10.13039\/501100002347","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Nat Mach Intell"],"abstract":"<jats:title>Abstract<\/jats:title>\n          <jats:p>Unsupervised learning has become an essential building block of artifical intelligence systems. The representations it produces, for example, in foundation models, are critical to a wide variety of downstream applications. It is therefore important to carefully examine unsupervised models to ensure not only that they produce accurate predictions on the available data but also that these accurate predictions do not arise from a Clever Hans (CH) effect. Here, using specially developed explainable artifical intelligence techniques and applying them to popular representation learning and anomaly detection models for image data, we show that CH effects are widespread in unsupervised learning. In particular, through use cases on medical and industrial inspection data, we demonstrate that CH effects systematically lead to significant performance loss of downstream models under plausible dataset shifts or reweighting of different data subgroups. Our empirical findings are enriched by theoretical insights, which point to inductive biases in the unsupervised learning machine as a primary source of CH effects. Overall, our work sheds light on unexplored risks associated with practical applications of unsupervised learning and suggests ways to systematically mitigate CH effects, thereby making unsupervised learning more robust.<\/jats:p>","DOI":"10.1038\/s42256-025-01000-2","type":"journal-article","created":{"date-parts":[[2025,3,17]],"date-time":"2025-03-17T10:02:52Z","timestamp":1742205772000},"page":"412-422","update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":18,"title":["Explainable AI reveals Clever Hans effects in unsupervised learning models"],"prefix":"10.1038","volume":"7","author":[{"given":"Jacob","family":"Kauffmann","sequence":"first","affiliation":[]},{"given":"Jonas","family":"Dippel","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9707-297X","authenticated-orcid":false,"given":"Lukas","family":"Ruff","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6283-3265","authenticated-orcid":false,"given":"Wojciech","family":"Samek","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3861-7685","authenticated-orcid":false,"given":"Klaus-Robert","family":"M\u00fcller","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7243-6186","authenticated-orcid":false,"given":"Gr\u00e9goire","family":"Montavon","sequence":"additional","affiliation":[]}],"member":"297","published-online":{"date-parts":[[2025,3,17]]},"reference":[{"key":"1000_CR1","unstructured":"Brown, T. B. et al. Language models are few-shot learners. In Advances in Neural Information Processing Systems, NeurIPS Vol. 33 (eds Larochelle, H. et al.) 1877\u20131901 (Curran Associates, 2020)."},{"key":"1000_CR2","doi-asserted-by":"crossref","first-page":"756","DOI":"10.1109\/JPROC.2021.3052449","volume":"109","author":"L Ruff","year":"2021","unstructured":"Ruff, L. et al. A unifying review of deep and shallow anomaly detection. Proc. IEEE 109, 756\u2013795 (2021).","journal-title":"Proc. IEEE"},{"key":"1000_CR3","doi-asserted-by":"crossref","first-page":"1346","DOI":"10.1038\/s41551-022-00914-1","volume":"6","author":"R Krishnan","year":"2022","unstructured":"Krishnan, R., Rajpurkar, P. & Topol, E. J. Self-supervised learning in medicine and healthcare. Nat. Biomed. Eng. 6, 1346\u20131352 (2022).","journal-title":"Nat. Biomed. Eng."},{"key":"1000_CR4","doi-asserted-by":"crossref","first-page":"2091","DOI":"10.1158\/0008-5472.CAN-08-2100","volume":"69","author":"A Li","year":"2009","unstructured":"Li, A. et al. Unsupervised analysis of transcriptomic profiles reveals six glioma subtypes. Cancer Res. 69, 2091\u20132099 (2009).","journal-title":"Cancer Res."},{"key":"1000_CR5","doi-asserted-by":"crossref","first-page":"20","DOI":"10.3389\/fgene.2019.00020","volume":"10","author":"L Jiang","year":"2019","unstructured":"Jiang, L., Xiao, Y., Ding, Y., Tang, J. & Guo, F. Discovering cancer subtypes via an accurate fusion strategy on multiple profile data. Front. Genet. 10, 20 (2019).","journal-title":"Front. Genet."},{"key":"1000_CR6","doi-asserted-by":"crossref","first-page":"eadj1719","DOI":"10.1126\/sciadv.adj1719","volume":"10","author":"O Eberle","year":"2024","unstructured":"Eberle, O. et al. Historical insights at scale: a corpus-wide machine learning analysis of early modern astronomic tables. Sci. Adv. 10, eadj1719 (2024).","journal-title":"Sci. Adv."},{"key":"1000_CR7","doi-asserted-by":"crossref","unstructured":"Rettig, L., Khayati, M., Cudr\u00e9-Mauroux, P. & Pi\u00f3rkowski, M. in Applied Data Science 289\u2013312 (Springer, 2019).","DOI":"10.1007\/978-3-030-11821-1_16"},{"key":"1000_CR8","doi-asserted-by":"crossref","unstructured":"Eskin, E., Arnold, A., Prerau, M. J., Portnoy, L. & Stolfo, S. J. in Applications of Data Mining in Computer Security, Advances in Information Security 77\u2013101 (Springer, 2002).","DOI":"10.1007\/978-1-4615-0953-0_4"},{"key":"1000_CR9","doi-asserted-by":"crossref","first-page":"1038","DOI":"10.1007\/s11263-020-01400-4","volume":"129","author":"P Bergmann","year":"2021","unstructured":"Bergmann, P., Batzner, K., Fauser, M., Sattlegger, D. & Steger, C. The MVTec anomaly detection dataset: a comprehensive real-world dataset for unsupervised anomaly detection. Int. J. Comput. Vis. 129, 1038\u20131059 (2021).","journal-title":"Int. J. Comput. Vis."},{"key":"1000_CR10","doi-asserted-by":"crossref","first-page":"109045","DOI":"10.1016\/j.cie.2023.109045","volume":"177","author":"J Zipfel","year":"2023","unstructured":"Zipfel, J. et al. Anomaly detection for industrial quality assurance: a comparative evaluation of unsupervised deep learning models. Comput. Ind. Eng. 177, 109045 (2023).","journal-title":"Comput. Ind. Eng."},{"key":"1000_CR11","doi-asserted-by":"publisher","unstructured":"Bommasani, R. et al. On the opportunities and risks of foundation models. Preprint at https:\/\/doi.org\/10.48550\/arXiv.2108.07258 (2021).","DOI":"10.48550\/arXiv.2108.07258"},{"key":"1000_CR12","unstructured":"Chen, T., Kornblith, S., Swersky, K., Norouzi, M. & Hinton, G. E. Big self-supervised models are strong semi-supervised learners. In Advances in Neural Information Processing Systems, NeurIPS Vol. 33 (eds Larochelle, H. et al.) 22243\u201322255 (Curran Associates, 2020)."},{"key":"1000_CR13","unstructured":"Radford, A. et al. Learning transferable visual models from natural language supervision. In ICML Proc. Machine Learning Research Vol. 139, 8748\u20138763 (2021)."},{"key":"1000_CR14","doi-asserted-by":"crossref","first-page":"259","DOI":"10.1038\/s41586-023-05881-4","volume":"616","author":"M Moor","year":"2023","unstructured":"Moor, M. et al. Foundation models for generalist medical artificial intelligence. Nature 616, 259\u2013265 (2023).","journal-title":"Nature"},{"key":"1000_CR15","doi-asserted-by":"publisher","unstructured":"Dippel, J. et al. RudolfV: a foundation model by pathologists for pathologists. Preprint at https:\/\/doi.org\/10.48550\/arXiv.2401.04079 (2024).","DOI":"10.48550\/arXiv.2401.04079"},{"key":"1000_CR16","doi-asserted-by":"crossref","DOI":"10.1038\/s41467-019-08987-4","volume":"10","author":"S Lapuschkin","year":"2019","unstructured":"Lapuschkin, S. et al. Unmasking Clever Hans predictors and assessing what machines really learn. Nat. Commun. 10, 1096 (2019).","journal-title":"Nat. Commun."},{"key":"1000_CR17","doi-asserted-by":"crossref","first-page":"665","DOI":"10.1038\/s42256-020-00257-z","volume":"2","author":"R Geirhos","year":"2020","unstructured":"Geirhos, R. et al. Shortcut learning in deep neural networks. Nat. Mach. Intell. 2, 665\u2013673 (2020).","journal-title":"Nat. Mach. Intell."},{"key":"1000_CR18","doi-asserted-by":"crossref","first-page":"476","DOI":"10.1038\/s42256-020-0212-3","volume":"2","author":"P Schramowski","year":"2020","unstructured":"Schramowski, P. et al. Making deep neural networks right for the right scientific reasons by interacting with their explanations. Nat. Mach. Intell. 2, 476\u2013486 (2020).","journal-title":"Nat. Mach. Intell."},{"key":"1000_CR19","doi-asserted-by":"crossref","first-page":"610","DOI":"10.1038\/s42256-021-00338-7","volume":"3","author":"AJ DeGrave","year":"2021","unstructured":"DeGrave, A. J., Janizek, J. D. & Lee, S.-I. Ai for radiographic COVID-19 detection selects shortcuts over signal. Nat. Mach. Intell. 3, 610\u2013619 (2021).","journal-title":"Nat. Mach. Intell."},{"key":"1000_CR20","doi-asserted-by":"crossref","first-page":"261","DOI":"10.1016\/j.inffus.2021.07.015","volume":"77","author":"CJ Anders","year":"2022","unstructured":"Anders, C. J. et al. Finding and removing Clever Hans: using explanation methods to debug and improve deep models. Inf. Fusion 77, 261\u2013295 (2022).","journal-title":"Inf. Fusion"},{"key":"1000_CR21","doi-asserted-by":"crossref","first-page":"102094","DOI":"10.1016\/j.inffus.2023.102094","volume":"103","author":"L Linhardt","year":"2024","unstructured":"Linhardt, L., M\u00fcller, K.-R. & Montavon, G. Preemptively pruning Clever-Hans strategies in deep neural networks. Inf. Fusion 103, 102094 (2024).","journal-title":"Inf. Fusion"},{"key":"1000_CR22","doi-asserted-by":"crossref","first-page":"1135","DOI":"10.1001\/jamadermatol.2019.1735","volume":"155","author":"JK Winkler","year":"2019","unstructured":"Winkler, J. K. et al. Association between surgical skin markings in dermoscopic images and diagnostic performance of a deep learning convolutional neural network for melanoma recognition. JAMA Dermatol. 155, 1135\u20131141 (2019).","journal-title":"JAMA Dermatol."},{"key":"1000_CR23","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1016\/j.dsp.2017.10.011","volume":"73","author":"G Montavon","year":"2018","unstructured":"Montavon, G., Samek, W. & M\u00fcller, K.-R. Methods for interpreting and understanding deep neural networks. Digit. Signal Process. 73, 1\u201315 (2018).","journal-title":"Digit. Signal Process."},{"key":"1000_CR24","doi-asserted-by":"crossref","first-page":"eaay7120","DOI":"10.1126\/scirobotics.aay7120","volume":"4","author":"D Gunning","year":"2019","unstructured":"Gunning, D. et al. XAI\u2014explainable artificial intelligence. Sci. Robot. 4, eaay7120 (2019).","journal-title":"Sci. Robot."},{"key":"1000_CR25","doi-asserted-by":"crossref","first-page":"82","DOI":"10.1016\/j.inffus.2019.12.012","volume":"58","author":"AB Arrieta","year":"2020","unstructured":"Arrieta, A. B. et al. Explainable artificial intelligence (XAI): concepts, taxonomies, opportunities and challenges toward responsible AI. Inf. Fusion 58, 82\u2013115 (2020).","journal-title":"Inf. Fusion"},{"key":"1000_CR26","doi-asserted-by":"crossref","first-page":"247","DOI":"10.1109\/JPROC.2021.3060483","volume":"109","author":"W Samek","year":"2021","unstructured":"Samek, W., Montavon, G., Lapuschkin, S., Anders, C. J. & M\u00fcller, K.-R. Explaining deep neural networks and beyond: a review of methods and applications. Proc. IEEE 109, 247\u2013278 (2021).","journal-title":"Proc. IEEE"},{"key":"1000_CR27","doi-asserted-by":"crossref","first-page":"541","DOI":"10.1146\/annurev-pathmechdis-051222-113147","volume":"19","author":"F Klauschen","year":"2024","unstructured":"Klauschen, F. et al. Toward explainable artificial intelligence for precision pathology. Annu. Rev. Pathol. 19, 541\u2013570 (2024).","journal-title":"Annu. Rev. Pathol."},{"key":"1000_CR28","doi-asserted-by":"crossref","first-page":"e0130140","DOI":"10.1371\/journal.pone.0130140","volume":"10","author":"S Bach","year":"2015","unstructured":"Bach, S. et al. On pixel-wise explanations for non-linear classifier decisions by layer-wise relevance propagation. PLoS ONE 10, e0130140 (2015).","journal-title":"PLoS ONE"},{"key":"1000_CR29","doi-asserted-by":"crossref","unstructured":"Montavon, G., Binder, A., Lapuschkin, S., Samek, W. & M\u00fcller, K.-R. in Lecture Notes in Computer Science Vol. 11700 (eds Samek, W. et al.) 193\u2013209 (Springer, 2019).","DOI":"10.1007\/978-3-030-28954-6_10"},{"key":"1000_CR30","doi-asserted-by":"crossref","first-page":"107198","DOI":"10.1016\/j.patcog.2020.107198","volume":"101","author":"J Kauffmann","year":"2020","unstructured":"Kauffmann, J., M\u00fcller, K.-R. & Montavon, G. Towards explaining anomalies: a deep Taylor decomposition of one-class models. Pattern Recognit. 101, 107198 (2020).","journal-title":"Pattern Recognit."},{"key":"1000_CR31","doi-asserted-by":"crossref","first-page":"1149","DOI":"10.1109\/TPAMI.2020.3020738","volume":"44","author":"O Eberle","year":"2022","unstructured":"Eberle, O. et al. Building and interpreting deep similarity models. IEEE Trans. Pattern Anal. Mach. Intell. 44, 1149\u20131161 (2022).","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"1000_CR32","doi-asserted-by":"crossref","first-page":"110309","DOI":"10.1016\/j.patcog.2024.110309","volume":"150","author":"J Vielhaben","year":"2024","unstructured":"Vielhaben, J., Lapuschkin, S., Montavon, G. & Samek, W. Explainable AI for time series via virtual inspection layers. Pattern Recognit. 150, 110309 (2024).","journal-title":"Pattern Recognit."},{"key":"1000_CR33","doi-asserted-by":"crossref","first-page":"7283","DOI":"10.1109\/TPAMI.2024.3388275","volume":"46","author":"P Chormai","year":"2024","unstructured":"Chormai, P., Herrmann, J., M\u00fcller, K.-R. & Montavon, G. Disentangled explanations of neural network predictions by finding relevant subspaces. IEEE Trans. Pattern Anal. Mach. Intell. 46, 7283\u20137299 (2024).","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"1000_CR34","unstructured":"Zhou, C. et al. LIMA: less is more for alignment. In Advances in Neural Information Processing Systems, NeurIPS Vol. 36 (eds Oh, A. et al.) 55006\u201355021 (Curran Associates, 2023)."},{"key":"1000_CR35","unstructured":"Muttenthaler, L., Dippel, J., Linhardt, L., Vandermeulen, R. A. & Kornblith, S. Human alignment of neural network representations. In Proc. International Conference on Learning Representations (ICLR) (OpenReview.net, 2023)."},{"key":"1000_CR36","doi-asserted-by":"crossref","unstructured":"Wang, X. et al. ChestX-Ray8: hospital-scale chest X-ray database and benchmarks on weakly-supervised classification and localization of common thorax diseases. In Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 3462\u20133471 (IEEE, 2017).","DOI":"10.1109\/CVPR.2017.369"},{"key":"1000_CR37","doi-asserted-by":"publisher","unstructured":"Cohen, J. P. et al. COVID-19 image data collection: prospective predictions are the future. Preprint at https:\/\/doi.org\/10.48550\/arXiv.2006.11988 (2020).","DOI":"10.48550\/arXiv.2006.11988"},{"key":"1000_CR38","doi-asserted-by":"crossref","unstructured":"Azizi, S. et al. Big self-supervised models advance medical image classification. In Proc. International Conference on Computer Vision (ICCV) 3458\u20133468 (IEEE, 2021).","DOI":"10.1109\/ICCV48922.2021.00346"},{"key":"1000_CR39","doi-asserted-by":"crossref","unstructured":"Eslami, S., Meinel, C. & de Melo, G. PubMedCLIP: how much does CLIP benefit visual question answering in the medical domain? In Proc. Findings of the Association for Computational Linguistics (EACL) 1151\u20131163 (Association for Computational Linguistics, 2023).","DOI":"10.18653\/v1\/2023.findings-eacl.88"},{"key":"1000_CR40","doi-asserted-by":"crossref","first-page":"74","DOI":"10.1038\/s41746-023-00811-0","volume":"6","author":"S-C Huang","year":"2023","unstructured":"Huang, S.-C. et al. Self-supervised learning for medical image classification: a systematic review and implementation guidelines. npj Digit. Med. 6, 74 (2023).","journal-title":"npj Digit. Med."},{"key":"1000_CR41","unstructured":"Chen, T., Kornblith, S., Norouzi, M. & Hinton, G. E. A simple framework for contrastive learning of visual representations. In ICML Proc. Machine Learning Research Vol. 119, 1597\u20131607 (PMLR, 2020)."},{"key":"1000_CR42","unstructured":"Zbontar, J., Jing, L., Misra, I., LeCun, Y. & Deny, S. Barlow twins: self-supervised learning via redundancy reduction. In ICML Proc. Machine Learning Research Vol. 139, 2310\u201312320 (PMLR, 2021)."},{"key":"1000_CR43","doi-asserted-by":"crossref","unstructured":"Deng, J. et al. Imagenet: a large-scale hierarchical image database. In Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 248\u2013255 (IEEE, 2009).","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"1000_CR44","unstructured":"Chen, T., Luo, C. & Li, L. Intriguing properties of contrastive losses. In Advances in Neural Information Processing Systems, NeurIPS Vol. 34 (eds Ranzato, M. et al.) 11834\u201311845 (Curran Associates, 2021)."},{"key":"1000_CR45","unstructured":"Robinson, J. et al. Can contrastive learning avoid shortcut solutions? In Advances in Neural Information Processing Systems, NeurIPS Vol. 34 (eds Ranzato, M. et al.) 4974\u20134986 (Curran Associates, 2021)."},{"key":"1000_CR46","unstructured":"Dippel, J., Vogler, S. & H\u00f6hne, J. Towards fine-grained visual representations by combining contrastive learning with image reconstruction and attention-weighted pooling. In ICML Workshop: Self-Supervised Learning for Reasoning and Perception (2021)."},{"key":"1000_CR47","doi-asserted-by":"crossref","unstructured":"Li, T. et al. Addressing feature suppression in unsupervised visual representations. In Proc. Winter Conference on Applications of Computer Vision (WACV) 1411\u20131420 (IEEE, 2023).","DOI":"10.1109\/WACV56688.2023.00146"},{"key":"1000_CR48","doi-asserted-by":"crossref","unstructured":"Roth, K. et al. Towards total recall in industrial anomaly detection. In IEEE\/CVF Conference on Computer Vision and Pattern Recognition 14298\u201314308 (IEEE, 2022).","DOI":"10.1109\/CVPR52688.2022.01392"},{"key":"1000_CR49","doi-asserted-by":"crossref","unstructured":"Batzner, K., Heckler, L. & K\u00f6nig, R. Efficientad: accurate visual anomaly detection at millisecond-level latencies. In Proc. Winter Conference on Applications of Computer Vision (WACV) 127\u2013137 (IEEE, 2024).","DOI":"10.1109\/WACV57701.2024.00020"},{"key":"1000_CR50","doi-asserted-by":"crossref","first-page":"1608","DOI":"10.1016\/j.neucom.2005.05.015","volume":"69","author":"S Harmeling","year":"2006","unstructured":"Harmeling, S., Dornhege, G., Tax, D., Meinecke, F. & M\u00fcller, K.-R. From outliers to prototypes: ordering data. Neurocomputing 69, 1608\u20131618 (2006).","journal-title":"Neurocomputing"},{"key":"1000_CR51","doi-asserted-by":"crossref","unstructured":"Aggarwal, C. C. Outlier Analysis (Springer, 2013).","DOI":"10.1007\/978-1-4614-6396-2"},{"key":"1000_CR52","doi-asserted-by":"crossref","first-page":"15:1\u201315:58","DOI":"10.1145\/1541880.1541882","volume":"41","author":"V Chandola","year":"2009","unstructured":"Chandola, V., Banerjee, A. & Kumar, V. Anomaly detection: a survey. ACM Comput. Surv. 41, 15:1\u201315:58 (2009).","journal-title":"ACM Comput. Surv."},{"key":"1000_CR53","doi-asserted-by":"crossref","first-page":"1065","DOI":"10.1214\/aoms\/1177704472","volume":"33","author":"E Parzen","year":"1962","unstructured":"Parzen, E. On estimation of a probability density function and mode. Ann. Math. Stat. 33, 1065\u20131076 (1962).","journal-title":"Ann. Math. Stat."},{"key":"1000_CR54","first-page":"2529","volume":"13","author":"J Kim","year":"2012","unstructured":"Kim, J. & Scott, C. D. Robust kernel density estimation. J. Mach. Learn. Res. 13, 2529\u20132565 (2012).","journal-title":"J. Mach. Learn. Res."},{"key":"1000_CR55","doi-asserted-by":"crossref","first-page":"1443","DOI":"10.1162\/089976601750264965","volume":"13","author":"B Sch\u00f6lkopf","year":"2001","unstructured":"Sch\u00f6lkopf, B., Platt, J. C., Shawe-Taylor, J., Smola, A. J. & Williamson, R. C. Estimating the support of a high-dimensional distribution. Neural Comput. 13, 1443\u20131471 (2001).","journal-title":"Neural Comput."},{"key":"1000_CR56","unstructured":"Montavon, G., Kauffmann, J. R., Samek, W. & M\u00fcller, K.-R. in Lecture Notes in Computer Science Vol. 13200 (eds Holzinger, A. et al.) 117\u2013138 (Springer, 2020)."},{"key":"1000_CR57","doi-asserted-by":"crossref","first-page":"767299","DOI":"10.3389\/fnbot.2021.767299","volume":"15","author":"Y Yu","year":"2022","unstructured":"Yu, Y., Qian, J. & Wu, Q. Visual saliency via multiscale analysis in frequency domain and its applications to ship detection in optical satellite images. Front. Neurorobot. 15, 767299 (2022).","journal-title":"Front. Neurorobot."},{"key":"1000_CR58","doi-asserted-by":"crossref","unstructured":"Parmar, G., Zhang, R. & Zhu, J. On aliased resizing and surprising subtleties in GAN evaluation. In Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 11400\u201311410 (IEEE, 2022).","DOI":"10.1109\/CVPR52688.2022.01112"},{"key":"1000_CR59","unstructured":"Kirichenko, P., Izmailov, P. & Wilson, A. G. Last layer re-training is sufficient for robustness to spurious correlations. In Proc. International Conference on Learning Representations (ICLR) (OpenReview.net, 2023)."},{"key":"1000_CR60","first-page":"985","volume":"8","author":"M Sugiyama","year":"2007","unstructured":"Sugiyama, M., Krauledat, M. & M\u00fcller, K.-R. Covariate shift adaptation by importance weighted cross validation. J. Mach. Learn. Res. 8, 985\u20131005 (2007).","journal-title":"J. Mach. Learn. Res."},{"key":"1000_CR61","doi-asserted-by":"crossref","unstructured":"Sugiyama, M. & Kawanabe, M. Machine Learning in Non-stationary Environments: Introduction to Covariate Shift Adaptation (MIT Press, 2012).","DOI":"10.7551\/mitpress\/9780262017091.001.0001"},{"key":"1000_CR62","unstructured":"Iwasawa, Y. & Matsuo, Y. Test-time classifier adjustment module for model-agnostic domain generalization. In Advances in Neural Information Processing Systems, NeurIPS Vol. 34 (eds Ranzato, M. et al.) 2427\u20132440 (Curran Associates, 2021)."},{"key":"1000_CR63","doi-asserted-by":"crossref","first-page":"2623","DOI":"10.1021\/acs.jcim.1c00160","volume":"61","author":"C Esposito","year":"2021","unstructured":"Esposito, C., Landrum, G. A., Schneider, N., Stiefl, N. & Riniker, S. Ghost: adjusting the decision threshold to handle imbalanced data in machine learning. J. Chem. Inf. Model. 61, 2623\u20132640 (2021).","journal-title":"J. Chem. Inf. Model."},{"key":"1000_CR64","doi-asserted-by":"crossref","unstructured":"Niven, T. & Kao, H. Probing neural network comprehension of natural language arguments. In Proc. Conference of the Association for Computational Linguistics (eds Korhonen, A. et al.) 4658\u20134664 (Association for Computational Linguistics, 2019).","DOI":"10.18653\/v1\/P19-1459"},{"key":"1000_CR65","first-page":"161","volume":"21","author":"B Heinzerling","year":"2020","unstructured":"Heinzerling, B. NLP\u2019s Clever Hans moment has arrived. J. Cogn. Sci. 21, 161\u2013170 (2020).","journal-title":"J. Cogn. Sci."},{"key":"1000_CR66","first-page":"1875","volume":"9","author":"ML Braun","year":"2008","unstructured":"Braun, M. L., Buhmann, J. M. & M\u00fcller, K.-R. On relevant dimensions in kernel feature spaces. J. Mach. Learn. Res. 9, 1875\u20131908 (2008).","journal-title":"J. Mach. Learn. Res."},{"key":"1000_CR67","unstructured":"Basri, R. et al. Frequency bias in neural networks for input of non-uniform density. In ICML Proc. Machine Learning Research Vol. 119, 685\u2013694 (PMLR, 2020)."},{"key":"1000_CR68","unstructured":"Fridovich-Keil, S., Lopes, R. G. & Roelofs, R. Spectral bias in practice: the role of function frequency in generalization. In Advances in Neural Information Processing Systems, NeurIPS Vol. 35 (eds Koyejo, S. et al.) (Curran Associates, 2022)."},{"key":"1000_CR69","doi-asserted-by":"crossref","unstructured":"Arras, L. et al. in Lecture Notes in Computer Science Vol. 11700 (eds Samek, W. et al.) 211\u2013238 (Springer, 2019).","DOI":"10.1007\/978-3-030-28954-6_11"},{"key":"1000_CR70","doi-asserted-by":"crossref","first-page":"7581","DOI":"10.1109\/TPAMI.2021.3115452","volume":"44","author":"T Schnake","year":"2022","unstructured":"Schnake, T. et al. Higher-order explanations of graph neural networks via relevant walks. IEEE Trans. Pattern Anal. Mach. Intell. 44, 7581\u20137596 (2022).","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"1000_CR71","unstructured":"Ali, A. et al. XAI for transformers: better explanations through conservative propagation. In ICML Proc. Machine Learning Research Vol. 162, 435\u2013451, (PMLR, 2022)."},{"key":"1000_CR72","unstructured":"Jafari, F. R., Montavon, G., Muller, K. R. & Eberle, O. MambaLRP: explaining selective state space sequence models. In Advances in Neural Information Processing Systems, NeurIPS Vol. 37 (eds Globerson, A. et al.) 118540\u2013118570 (Curran Associates, 2024)."},{"key":"1000_CR73","doi-asserted-by":"crossref","first-page":"1991","DOI":"10.1109\/ACCESS.2018.2886457","volume":"7","author":"M Munir","year":"2019","unstructured":"Munir, M., Siddiqui, S. A., Dengel, A. & Ahmed, S. DeepAnT: a deep learning approach for unsupervised anomaly detection in time series. IEEE Access 7, 1991\u20132005 (2019).","journal-title":"IEEE Access"},{"key":"1000_CR74","unstructured":"Devlin, J., Chang, M., Lee, K. & Toutanova, K. BERT: pre-training of deep bidirectional transformers for language understanding. In Proc. Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) (eds Burstein, J. et al.) 4171\u20134186 (Association for Computational Linguistics, 2019)."},{"key":"1000_CR75","doi-asserted-by":"crossref","unstructured":"Pelka, O., Koitka, S., R\u00fcckert, J., Nensa, F. & Friedrich, C. M. in Lecture Notes in Computer Science Vol. 11043 (eds Stoyanov, D. et al.) 180\u2013189 (Springer, 2018).","DOI":"10.1007\/978-3-030-01364-6_20"},{"key":"1000_CR76","unstructured":"PubMedCLIP Hugging Face https:\/\/huggingface.co\/sarahESL\/PubMedCLIP (2024)."},{"key":"1000_CR77","unstructured":"openaiCLIP GitHub https:\/\/github.com\/openai\/CLIP (2024)."},{"key":"1000_CR78","unstructured":"Facebook Research. barlowtwins GitHub https:\/\/github.com\/facebookresearch\/barlowtwins (2022)."},{"key":"1000_CR79","unstructured":"CXR8 NIHCC https:\/\/nihcc.app.box.com\/v\/ChestXray-NIHCC (2017)."},{"key":"1000_CR80","unstructured":"ieee8023 COVID-chestxray-dataset GitHub https:\/\/github.com\/ieee8023\/covid-chestxray-dataset (2020)."},{"key":"1000_CR81","doi-asserted-by":"crossref","first-page":"38:1\u201338:38","DOI":"10.1145\/3439950","volume":"54","author":"G Pang","year":"2022","unstructured":"Pang, G., Shen, C., Cao, L. & van den Hengel, A. Deep learning for anomaly detection: a review. ACM Comput. Surv. 54, 38:1\u201338:38 (2022).","journal-title":"ACM Comput. Surv."},{"key":"1000_CR82","doi-asserted-by":"crossref","unstructured":"Rippel, O., Mertens, P. & Merhof, D. Modeling the distribution of normal data in pre-trained deep features for anomaly detection. In ICPR 6726\u20136733 (IEEE, 2020).","DOI":"10.1109\/ICPR48806.2021.9412109"},{"key":"1000_CR83","doi-asserted-by":"crossref","first-page":"S63","DOI":"10.1121\/1.2016299","volume":"62","author":"F Jelinek","year":"2005","unstructured":"Jelinek, F., Mercer, R. L., Bahl, L. R. & Baker, J. K. Perplexity\u2014a measure of the difficulty of speech recognition tasks. J. Acoust. Soc. Am. 62, S63\u2013S63 (2005).","journal-title":"J. Acoust. Soc. Am."},{"key":"1000_CR84","unstructured":"Amazon Science. patchcore-inspection GitHub https:\/\/github.com\/amazon-science\/patchcore-inspection (2022)."},{"key":"1000_CR85","unstructured":"Bradski, G. The OpenCV library. Dr. Dobb\u2019s Journal of Software Tools 120, 122\u2013125 (2000)."},{"key":"1000_CR86","unstructured":"Torchvision: PyTorch\u2019s computer vision library https:\/\/pytorch.org\/vision (2016)."},{"key":"1000_CR87","doi-asserted-by":"crossref","unstructured":"Samek, W., Montavon, G., Vedaldi, A., Hansen, L. K. & M\u00fcller, K.-R. Explainable AI: Interpreting, Explaining And Visualizing Deep Learning Vol. 11700 (Springer, 2019).","DOI":"10.1007\/978-3-030-28954-6"},{"key":"1000_CR88","unstructured":"zennit. GitHub https:\/\/github.com\/chr5tphr\/zennit (2021)."},{"key":"1000_CR89","doi-asserted-by":"publisher","unstructured":"Kauffmann, J. et al. Explainable AI reveals clever hans effects in unsupervised learning models: code. Zenodo https:\/\/doi.org\/10.5281\/zenodo.14186119 (2024).","DOI":"10.5281\/zenodo.14186119"},{"key":"1000_CR90","unstructured":"Ml-workgroup. COVID-19 image repository. GitHub https:\/\/github.com\/ml-workgroup\/covid-19-image-repository (2020)."}],"container-title":["Nature Machine Intelligence"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.nature.com\/articles\/s42256-025-01000-2.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s42256-025-01000-2","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s42256-025-01000-2.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,3,24]],"date-time":"2025-03-24T23:29:30Z","timestamp":1742858970000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.nature.com\/articles\/s42256-025-01000-2"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,3,17]]},"references-count":90,"journal-issue":{"issue":"3","published-online":{"date-parts":[[2025,3]]}},"alternative-id":["1000"],"URL":"https:\/\/doi.org\/10.1038\/s42256-025-01000-2","relation":{},"ISSN":["2522-5839"],"issn-type":[{"value":"2522-5839","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,3,17]]},"assertion":[{"value":"27 July 2024","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"19 December 2024","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"17 March 2025","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"The authors declare no competing interests.","order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}]}}