{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,12,27]],"date-time":"2025-12-27T15:08:49Z","timestamp":1766848129651,"version":"3.37.3"},"reference-count":35,"publisher":"Springer Science and Business Media LLC","issue":"9","license":[{"start":{"date-parts":[[2023,3,14]],"date-time":"2023-03-14T00:00:00Z","timestamp":1678752000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2023,3,14]],"date-time":"2023-03-14T00:00:00Z","timestamp":1678752000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"DOI":"10.13039\/100019180","name":"HORIZON EUROPE European Research Council","doi-asserted-by":"publisher","award":["761758","761758"],"award-info":[{"award-number":["761758","761758"]}],"id":[{"id":"10.13039\/100019180","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Mach Learn"],"published-print":{"date-parts":[[2023,9]]},"abstract":"<jats:title>Abstract<\/jats:title><jats:p>Recent work has shown learning systems can use logical background knowledge to compensate for a lack of labeled training data. Many methods work by creating a loss function that encodes this knowledge. However, often the logic is discarded after training, even if it is still helpful at test time. Instead, we ensure neural network predictions satisfy the knowledge by refining the predictions with an extra computation step. We introduce differentiable <jats:italic>refinement functions<\/jats:italic> that find a corrected prediction close to the original prediction. We study how to effectively and efficiently compute these refinement functions. Using a new algorithm called iterative local refinement (ILR), we combine refinement functions to find refined predictions for logical formulas of any complexity. ILR finds refinements on complex SAT formulas in significantly fewer iterations and frequently finds solutions where gradient descent can not. Finally, ILR produces competitive results in the MNIST addition task.<\/jats:p>","DOI":"10.1007\/s10994-023-06310-3","type":"journal-article","created":{"date-parts":[[2023,3,14]],"date-time":"2023-03-14T22:02:30Z","timestamp":1678831350000},"page":"3293-3331","update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":8,"title":["Refining neural network predictions using background knowledge"],"prefix":"10.1007","volume":"112","author":[{"given":"Alessandro","family":"Daniele","sequence":"first","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5502-4817","authenticated-orcid":false,"given":"Emile","family":"van Krieken","sequence":"additional","affiliation":[]},{"given":"Luciano","family":"Serafini","sequence":"additional","affiliation":[]},{"given":"Frank","family":"van Harmelen","sequence":"additional","affiliation":[]}],"member":"297","published-online":{"date-parts":[[2023,3,14]]},"reference":[{"key":"6310_CR1","unstructured":"Ahmed, K., Teso, S., Chang, K.-W., den Broeck, G.\u00a0V., & Vergari, A. (2022) Semantic probabilistic layers for neuro-symbolic learning. CoRR, arXiv:2206.00426."},{"key":"6310_CR2","doi-asserted-by":"publisher","unstructured":"Alsina, C. (1984) On Schur-Concave t-norms and triangle functions. In: W.\u00a0Walter, editor, General Inequalities 4: In Memoriam Edwin F. Beckenbach 4th International Conference on General Inequalities, Oberwolfach, May 8\u201314, 1983, pages 241\u2013248. Birkh\u00e4user, Basel, ISBN 978-3-0348-6259-2. https:\/\/doi.org\/10.1007\/978-3-0348-6259-2_22.","DOI":"10.1007\/978-3-0348-6259-2_22"},{"key":"6310_CR3","doi-asserted-by":"publisher","DOI":"10.1016\/j.artint.2021.103649","volume":"303","author":"S Badreddine","year":"2022","unstructured":"Badreddine, S., d\u2019Avila Garcez, A., Serafini, L., & Spranger, M. (2022). Logic tensor networks. Artificial Intelligence, 303, 103649.","journal-title":"Artificial Intelligence"},{"key":"6310_CR4","doi-asserted-by":"crossref","unstructured":"Calvo, T., Koles\u00e1rov\u00e1, A., Komorn\u00edkov\u00e1, M., & Mesiar, R. (2002) Aggregation operators: Properties, classes and construction methods. In T.\u00a0Calvo, G.\u00a0Mayor, and R.\u00a0Mesiar, editors, Aggregation Operators: New Trends and Applications, pages 3\u2013104. Physica-Verlag HD, Heidelberg, ISBN 978-3-7908-1787-4.","DOI":"10.1007\/978-3-7908-1787-4_1"},{"key":"6310_CR5","unstructured":"Chowdhery, A., Narang, S., Devlin, J., Bosma, M., Mishra, G., Roberts, A., Barham, P., Chung, H.\u00a0W., Sutton, C., Gehrmann, S., Schuh, P., Shi, K., Tsvyashchenko, S., Maynez, J., Rao, A., Barnes, P., Tay, Y., Shazeer, N., Prabhakaran, V., Reif, E., Du, N., Hutchinson, B., Pope, R., Bradbury, J., Austin, J., Isard, M., Gur-Ari, G., Yin, P., Duke, T., Levskaya, A., Ghemawat, S., Dev, S., Michalewski, H., Garcia, X., Misra, V., Robinson, K., Fedus, L., Zhou, D., Ippolito, D., Luan, D., Lim, H., Zoph, B., Spiridonov, A., Sepassi, R., Dohan, D., Agrawal, S., Omernick, M., Dai, A.\u00a0M., Pillai, T.\u00a0S., Pellat, M., Lewkowycz, A., Moreira, E., Child, R., Polozov, O., Lee, K., Zhou, Z., Wang, X., Saeta, B., Diaz, M., Firat, O., Catasta, M., Wei, J., Meier-Hellstern, K., Eck, D., Dean, J., Petrov, S., & Fiedel, N. (2022) PaLM: Scaling Language modeling with pathways. arXiv:2204.02311."},{"key":"6310_CR6","doi-asserted-by":"publisher","unstructured":"Clarke, F.\u00a0H., Stern, R.\u00a0J., & Wolenski, P.\u00a0R. (1993) Subgradient Criteria for Monotonicity, The Lipschitz Condition, and Convexity. Canadian Journal of Mathematics, 45(6):1167\u20131183, Dec. 1993. ISSN 0008-414X, 1496-4279. https:\/\/doi.org\/10.4153\/CJM-1993-065-x.","DOI":"10.4153\/CJM-1993-065-x"},{"key":"6310_CR7","doi-asserted-by":"crossref","unstructured":"Daniele, A., & Serafini, L. (2019) Knowledge enhanced neural networks. In A.\u00a0C. Nayak and A.\u00a0Sharma, editors, PRICAI 2019: Trends in Artificial Intelligence, pages 542\u2013554, Cham. Springer International Publishing. ISBN 978-3-030-29908-8.","DOI":"10.1007\/978-3-030-29908-8_43"},{"key":"6310_CR8","unstructured":"Daniele, A., & Serafini, L. (2022) Knowledge enhanced neural networks for relational domains. arXiv preprint arXiv:2205.15762."},{"key":"6310_CR9","doi-asserted-by":"publisher","first-page":"143","DOI":"10.1016\/j.artint.2015.08.011","volume":"244","author":"M Diligenti","year":"2017","unstructured":"Diligenti, M., Gori, M., & Sacca, C. (2017). Semantic-based regularization for learning and inference. Artificial Intelligence, 244, 143\u2013165.","journal-title":"Artificial Intelligence"},{"key":"6310_CR10","doi-asserted-by":"crossref","unstructured":"Donadello, I., Serafini, L., & d\u2019Avila Garcez, A. (2017) Logic tensor networks for semantic image interpretation. In IJCAI International joint conference on artificial intelligence, pp. 1596\u20141602.","DOI":"10.24963\/ijcai.2017\/221"},{"key":"6310_CR11","unstructured":"Dragone, P., Teso, S., & Passerini, A. (2021) Neuro-symbolic constraint programming for structured prediction. In: A.\u00a0S. d\u2019Avila Garcez and E.\u00a0Jim\u00e9nez-Ruiz, editors, Proceedings of the 15th international workshop on neural-symbolic learning and reasoning as part of the 1st international joint conference on learning & reasoning (IJCLR 2021), Virtual conference, October 25-27, 2021, volume 2986 of CEUR workshop proceedings, pages 6\u201314. CEUR-WS.org."},{"key":"6310_CR12","unstructured":"Fischer, M., Balunovic, M., Drachsler-Cohen, D., Gehr, T., Zhang, C., & Vechev, M.\u00a0T.(2019) DL2: Training and querying neural networks with logic. In: K.\u00a0Chaudhuri and R.\u00a0Salakhutdinov, editors, Proceedings of the 36th international conference on machine learning, ICML 2019, 9-15, Long Beach, California, USA, volume\u00a097 of Proceedings of machine learning research, pp. 1931\u20131941. PMLR."},{"issue":"7","key":"6310_CR13","doi-asserted-by":"publisher","first-page":"1407","DOI":"10.1109\/TFUZZ.2018.2879627","volume":"27","author":"F Giannini","year":"2019","unstructured":"Giannini, F., Diligenti, M., Gori, M., & Maggini, M. (2019). On a convex logic fragment for learning and reasoning. IEEE Transactions on Fuzzy Systems, 27(7), 1407\u20131416. https:\/\/doi.org\/10.1109\/TFUZZ.2018.2879627","journal-title":"IEEE Transactions on Fuzzy Systems"},{"key":"6310_CR14","doi-asserted-by":"publisher","first-page":"759","DOI":"10.1613\/jair.1.12850","volume":"72","author":"E Giunchiglia","year":"2021","unstructured":"Giunchiglia, E., & Lukasiewicz, T. (2021). Multi-label classification neural networks with hard logical constraints. Journal of Artificial Intelligence Research, 72, 759\u2013818. https:\/\/doi.org\/10.1613\/jair.1.12850","journal-title":"Journal of Artificial Intelligence Research"},{"key":"6310_CR15","doi-asserted-by":"crossref","unstructured":"Giunchiglia, E., Stoian, M., Khan, S., Cuzzolin, F., & Lukasiewicz, T. (2022a) ROAD-R: The autonomous driving dataset with logical requirements. June 2022a.","DOI":"10.1007\/s10994-023-06322-z"},{"key":"6310_CR16","doi-asserted-by":"publisher","unstructured":"Giunchiglia, E., Stoian, M.\u00a0C., & Lukasiewicz, T. (2022) Deep learning with logical constraints. In L.\u00a0D. Raedt, editor, Proceedings of the thirty-first international joint conference on artificial intelligence, IJCAI 2022, Vienna, Austria, 23-29 , pp. 5478\u20135485. ijcai.org, 2022b. https:\/\/doi.org\/10.24963\/ijcai.2022\/767.","DOI":"10.24963\/ijcai.2022\/767"},{"key":"6310_CR17","doi-asserted-by":"publisher","unstructured":"Hoernle, N., Karampatsis, R.\u00a0M., Belle, V., & Gal, K. (2022) MultiplexNet: Towards fully satisfied logical constraints in neural networks. In: Proceedings of the AAAI conference on artificial intelligence, 36(5):5700\u20135709. ISSN 2374-3468, 2159-5399. https:\/\/doi.org\/10.1609\/aaai.v36i5.20512.","DOI":"10.1609\/aaai.v36i5.20512"},{"key":"6310_CR18","unstructured":"Hoos, H.\u00a0H.(2000) SATLIB : An online resource for research on SAT. pp. 1\u201312."},{"key":"6310_CR19","unstructured":"Inc, W.\u00a0R. (2019) Mathematica, Version 12.0. 2019. Champaign, IL."},{"key":"6310_CR20","volume-title":"Fuzzy Implications","author":"B Jayaram","year":"2008","unstructured":"Jayaram, B., & Baczynski, M. (2008). Fuzzy Implications (Vol. 231). Berlin: Springer."},{"key":"6310_CR21","unstructured":"Kingma, D.\u00a0P., & Ba, J. (2015) Adam: A method for stochastic optimization. arXiv:1412.6980 [cs], Jan. 2017. Comment: Published as a conference paper at the 3rd International Conference for Learning Representations, San Diego."},{"key":"6310_CR22","doi-asserted-by":"publisher","unstructured":"Klement, E.-P, Mesiar, R., & Pap, E. (2000) Triangular Norms, volume\u00a08 of Trends in Logic. Springer. ISBN 978-90-481-5507-1. https:\/\/doi.org\/10.1007\/978-94-015-9540-7.","DOI":"10.1007\/978-94-015-9540-7"},{"key":"6310_CR23","doi-asserted-by":"publisher","unstructured":"Klement, E.\u00a0P, Mesiar, R., & Pap, E. (2004) Triangular norms. Position paper II: General constructions and parameterized families. Fuzzy Sets and Systems, 145(3):411\u2013438. ISSN 01650114. https:\/\/doi.org\/10.1016\/S0165-0114(03)00327-0.","DOI":"10.1016\/S0165-0114(03)00327-0"},{"key":"6310_CR24","unstructured":"LeCun, Y., & Cortes, C. (2010) MNIST handwritten digit database."},{"key":"6310_CR25","volume-title":"Advances in Neural Information Processing Systems 31: Annual Conference on Neural Information Processing Systems 2018, NeurIPS 2018, 3\u20138 December 2018","author":"R Manhaeve","year":"2018","unstructured":"Manhaeve, R., Duman\u010di\u0107, S., Kimmig, A., Demeester, T., & De Raedt, L. (2018). DeepProbLog: Neural probabilistic logic programming. In S. Bengio, H. M. Wallach, H. Larochelle, K. Grauman, N. Cesa-Bianchi, & R. Garnett (Eds.), Advances in Neural Information Processing Systems 31: Annual Conference on Neural Information Processing Systems 2018, NeurIPS 2018, 3\u20138 December 2018. Canada: Montr\u00e9al."},{"key":"6310_CR26","doi-asserted-by":"publisher","unstructured":"Marshall, A.\u00a0W., Olkin, I., & Arnold, B.\u00a0C. (2011) Schur-convex functions. In A.\u00a0W. Marshall, I.\u00a0Olkin, and B.\u00a0C. Arnold, editors, Inequalities: Theory of majorization and its applications, pp. 79\u2013154. Springer, New York, NY. ISBN 978-0-387-68276-1. https:\/\/doi.org\/10.1007\/978-0-387-68276-1_3.","DOI":"10.1007\/978-0-387-68276-1_3"},{"key":"6310_CR27","unstructured":"Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., & Chen, M. (2022) Hierarchical text-conditional image generation with CLIP Latents. arXiv:2204.06125."},{"key":"6310_CR28","doi-asserted-by":"publisher","unstructured":"T.\u00a0J. U. o. N.\u00a0M. Ross. Fuzzy logic with engineering applications. 2010. ISBN 978-0-470-74376-8. https:\/\/doi.org\/10.1002\/9781119994374.","DOI":"10.1002\/9781119994374"},{"key":"6310_CR29","doi-asserted-by":"publisher","unstructured":"Roychowdhury, S., Diligenti, M., & Gori, M. (2021) Regularizing deep networks with prior knowledge: A constraint-based approach. Knowledge-Based Systems, 222:106989. ISSN 0950-7051. https:\/\/doi.org\/10.1016\/j.knosys.2021.106989.","DOI":"10.1016\/j.knosys.2021.106989"},{"issue":"1","key":"6310_CR30","first-page":"50","volume":"155","author":"A Taka\u010di","year":"2005","unstructured":"Taka\u010di, A. (2005). Schur-concave triangular norms: Characterization and application in pFCSP. Fuzzy Sets and Systems. An International Journal in Information Science and Engineering, 155(1), 50\u201364.","journal-title":"Fuzzy Sets and Systems. An International Journal in Information Science and Engineering"},{"key":"6310_CR31","doi-asserted-by":"publisher","unstructured":"Van\u00a0Dyke, H.\u00a0A., Vixie, K.\u00a0R., & Asaki, T.\u00a0J. (2013) Cone Monotonicity: Structure Theorem, Properties, and Comparisons to Other Notions of Monotonicity. Abstract and Applied Analysis, 2013:1\u20138, 2013. ISSN 1085-3375, 1687-0409. https:\/\/doi.org\/10.1155\/2013\/134751.","DOI":"10.1155\/2013\/134751"},{"key":"6310_CR32","doi-asserted-by":"publisher","unstructured":"van Krieken, E., Acar, E., & van Harmelen, F. (2022) Analyzing differentiable fuzzy logic operators. Artificial Intelligence, 302:103602. ISSN 0004-3702. https:\/\/doi.org\/10.1016\/j.artint.2021.103602.","DOI":"10.1016\/j.artint.2021.103602"},{"key":"6310_CR33","unstructured":"Wang, P.-W., Donti, P.\u00a0L., Wilder, B., & Kolter, J.\u00a0Z. (2019) SATNet: Bridging deep learning and logical reasoning using a differentiable satisfiability solver. In K.\u00a0Chaudhuri and R.\u00a0Salakhutdinov, editors, Proceedings of the 36th international conference on machine learning, ICML 2019, 9-15 June 2019, Long Beach, California, USA, volume\u00a097 of Proceedings of Machine Learning Research, pages 6545\u20136554. PMLR."},{"key":"6310_CR34","unstructured":"Xu, J., Zhang, Z., Friedman, T., Liang, Y., & den Broeck, G. (2018) A semantic loss function for deep learning with symbolic knowledge. In J.\u00a0Dy and A.\u00a0Krause, editors, Proceedings of the 35th international conference on machine learning, volume\u00a080, pages 5502\u20135511, Stockholmsm\u00e4ssan, Stockholm Sweden, PMLR."},{"key":"6310_CR35","unstructured":"Yang, Z., Lee, J., & Park, C. (2022) Injecting Logical Constraints into Neural Networks via Straight-Through Estimators. In K.\u00a0Chaudhuri, S.\u00a0Jegelka, L.\u00a0Song, C.\u00a0Szepesv\u00e1ri, G.\u00a0Niu, and S.\u00a0Sabato, editors, International conference on machine learning, ICML 2022, 17-23 , Baltimore, Maryland, USA, volume 162 of Proceedings of machine learning research, pp. 25096\u201325122. PMLR, (2022)."}],"container-title":["Machine Learning"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s10994-023-06310-3.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s10994-023-06310-3\/fulltext.html","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s10994-023-06310-3.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,8,10]],"date-time":"2023-08-10T17:06:59Z","timestamp":1691687219000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s10994-023-06310-3"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,3,14]]},"references-count":35,"journal-issue":{"issue":"9","published-print":{"date-parts":[[2023,9]]}},"alternative-id":["6310"],"URL":"https:\/\/doi.org\/10.1007\/s10994-023-06310-3","relation":{},"ISSN":["0885-6125","1573-0565"],"issn-type":[{"type":"print","value":"0885-6125"},{"type":"electronic","value":"1573-0565"}],"subject":[],"published":{"date-parts":[[2023,3,14]]},"assertion":[{"value":"8 June 2022","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"23 January 2023","order":2,"name":"revised","label":"Revised","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"27 January 2023","order":3,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"14 March 2023","order":4,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"The authors have no conflicts of interest to declare that are relevant to the content of this article","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Conflict of interest"}},{"value":"We declare that our manuscript follows the ethics rules provided in .","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Ethical approval"}},{"value":"Not applicable","order":4,"name":"Ethics","group":{"name":"EthicsHeading","label":"Consent to participate"}},{"value":"Not applicable","order":5,"name":"Ethics","group":{"name":"EthicsHeading","label":"Consent for publication"}}]}}