{"doi":"10.52202/075280-2050","month":"12","article_processing_charge":"No","publication_identifier":{"eissn":["1049-5258"]},"external_id":{"arxiv":["2310.13387"]},"_id":"14954","title":"Assumption violations in causal discovery and the robustness of score matching","year":"2023","arxiv":1,"intvolume":"        36","status":"public","researchdata_availability":"no","publisher":"Neural Information Processing Systems Foundation","author":[{"full_name":"Montagna, Francesco","first_name":"Francesco","last_name":"Montagna"},{"full_name":"Mastakouri, Atalanti A.","first_name":"Atalanti A.","last_name":"Mastakouri"},{"full_name":"Eulig, Elias","last_name":"Eulig","first_name":"Elias"},{"first_name":"Nicoletta","last_name":"Noceti","full_name":"Noceti, Nicoletta"},{"full_name":"Rosasco, Lorenzo","first_name":"Lorenzo","last_name":"Rosasco"},{"full_name":"Janzing, Dominik","last_name":"Janzing","first_name":"Dominik"},{"full_name":"Aragam, Bryon","last_name":"Aragam","first_name":"Bryon"},{"last_name":"Locatello","first_name":"Francesco","full_name":"Locatello, Francesco","orcid":"0000-0002-4850-0683","id":"26cfd52f-2483-11ee-8040-88983bcc06d4"}],"date_updated":"2026-08-13T07:51:13Z","citation":{"chicago":"Montagna, Francesco, Atalanti A. Mastakouri, Elias Eulig, Nicoletta Noceti, Lorenzo Rosasco, Dominik Janzing, Bryon Aragam, and Francesco Locatello. “Assumption Violations in Causal Discovery and the Robustness of Score Matching.” In <i>37th Conference on Neural Information Processing Systems</i>, Vol. 36. Neural Information Processing Systems Foundation, 2023. <a href=\"https://doi.org/10.52202/075280-2050\">https://doi.org/10.52202/075280-2050</a>.","ama":"Montagna F, Mastakouri AA, Eulig E, et al. Assumption violations in causal discovery and the robustness of score matching. In: <i>37th Conference on Neural Information Processing Systems</i>. Vol 36. Neural Information Processing Systems Foundation; 2023. doi:<a href=\"https://doi.org/10.52202/075280-2050\">10.52202/075280-2050</a>","ista":"Montagna F, Mastakouri AA, Eulig E, Noceti N, Rosasco L, Janzing D, Aragam B, Locatello F. 2023. Assumption violations in causal discovery and the robustness of score matching. 37th Conference on Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information Processing Systems, vol. 36.","apa":"Montagna, F., Mastakouri, A. A., Eulig, E., Noceti, N., Rosasco, L., Janzing, D., … Locatello, F. (2023). Assumption violations in causal discovery and the robustness of score matching. In <i>37th Conference on Neural Information Processing Systems</i> (Vol. 36). New Orleans, LO, United States: Neural Information Processing Systems Foundation. <a href=\"https://doi.org/10.52202/075280-2050\">https://doi.org/10.52202/075280-2050</a>","short":"F. Montagna, A.A. Mastakouri, E. Eulig, N. Noceti, L. Rosasco, D. Janzing, B. Aragam, F. Locatello, in:, 37th Conference on Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2023.","ieee":"F. Montagna <i>et al.</i>, “Assumption violations in causal discovery and the robustness of score matching,” in <i>37th Conference on Neural Information Processing Systems</i>, New Orleans, LO, United States, 2023, vol. 36.","mla":"Montagna, Francesco, et al. “Assumption Violations in Causal Discovery and the Robustness of Score Matching.” <i>37th Conference on Neural Information Processing Systems</i>, vol. 36, Neural Information Processing Systems Foundation, 2023, doi:<a href=\"https://doi.org/10.52202/075280-2050\">10.52202/075280-2050</a>."},"publication":"37th Conference on Neural Information Processing Systems","supplementarymaterial":"yes","das_tickbox":"0","file":[{"date_created":"2026-08-13T07:49:38Z","content_type":"application/pdf","file_id":"22701","success":1,"access_level":"open_access","date_updated":"2026-08-13T07:49:38Z","creator":"dernst","checksum":"71d38402b4edef4c084b39f3e2f04526","relation":"main_file","file_size":7640984,"file_name":"2023_Neurips_Montagna.pdf"}],"oa_version":"Published Version","date_published":"2023-12-20T00:00:00Z","has_accepted_license":"1","department":[{"_id":"FrLo"}],"user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","ddc":["000"],"file_date_updated":"2026-08-13T07:49:38Z","type":"conference","publication_status":"published","volume":36,"language":[{"iso":"eng"}],"OA_type":"gold","alternative_title":["Advances in Neural Information Processing Systems"],"OA_place":"publisher","quality_controlled":"1","day":"20","fulldoi":"https://doi.org/10.52202/075280-2050","date_created":"2024-02-07T15:11:56Z","acknowledgement":"We thank Kun Zhang and Carl-Johann Simon-Gabriel for the insightful discussions. This work\r\nhas been supported by AFOSR, grant n. FA8655-20-1-7035. FM is supported by Programma\r\nOperativo Nazionale ricerca e innovazione 2014-2020. FM partially contributed to this work during an internship at Amazon Web Services with FL. FL partially contributed while at AWS.","conference":{"location":"New Orleans, LO, United States","start_date":"2023-12-12","end_date":"2023-12-16","name":"NeurIPS: Neural Information Processing Systems"},"oa":1,"abstract":[{"text":"When domain knowledge is limited and experimentation is restricted by ethical,\r\nfinancial, or time constraints, practitioners turn to observational causal discovery\r\nmethods to recover the causal structure, exploiting the statistical properties of their\r\ndata. Because causal discovery without further assumptions is an ill-posed problem,\r\neach algorithm comes with its own set of usually untestable assumptions, some\r\nof which are hard to meet in real datasets. Motivated by these considerations, this\r\npaper extensively benchmarks the empirical performance of recent causal discovery\r\nmethods on observational iid data generated under different background conditions,\r\nallowing for violations of the critical assumptions required by each selected approach. Our experimental findings show that score matching-based methods demonstrate surprising performance in the false positive and false negative rate of the\r\ninferred graph in these challenging scenarios, and we provide theoretical insights\r\ninto their performance. This work is also the first effort to benchmark the stability of\r\ncausal discovery algorithms with respect to the values of their hyperparameters. Finally, we hope this paper will set a new standard for the evaluation of causal discovery methods and can serve as an accessible entry point for practitioners interested\r\nin the field, highlighting the empirical implications of different algorithm choices.","lang":"eng"}]}