[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-127798-en":3,"doc-seo-127798-105":30,"detail-sidebar-cat-0-en-105":92},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},127798,1099523882367,"Hazel","https://ap-avatar.wpscdn.com/davatar_9964176cb1d06d4a9deccf72a44ae3dc",8,"Research & Report","Interpretable Neural-Symbolic Concept Reasoning - Deep Concept Reasoner (DCR)","Deep learning achieves high accuracy, but its opaque decision mechanisms limit human trust. Concept-based models improve interpretability by learning from human-understandable concepts; however, state-of-the-art approaches often use high-dimensional concept embeddings whose dimensions lack clear semantic meaning. The Deep Concept Reasoner (DCR) builds differentiable syntactic rule structures from concept embeddings, executes them over meaningful fuzzy concept truth degrees, and yields semantically consistent predictions. Experiments show up to +25% gains, logic rule discovery without concept supervision, and useful counterfactual generation.","Interpretable Neural-Symbolic Concept Reasoning  \nPietro Barbiero * 1 Gabriele Ciravegna * 2 Francesco Giannini * 3 Mateo Espinosa Zarlenga 1 Lucie Charlotte Magister 1 Alberto Tonda 4 Pietro Li 1 Frederic Precioso 2 Mateja Jamnik 1  \nGiuseppe Marra * 5  \nAbstract  \nDeep learning methods are highly accurate, yet their opaque decision process prevents them from earning full human trust. Concept-based models aim to address this issue by learning tasks based on a set of human-understandable concepts.  \nHowever, state-of-the-art concept-based models rely on high-dimensional concept embedding representations which lack a clear semantic meaning, thus questioning the interpretability of their decision process. To overcome this limitation, we propose the Deep Concept Reasoner (DCR), the first interpretable concept-based model that builds upon concept embeddings. In DCR, neural networks do not make task predictions directly, but they build syntactic rule structures using concept embeddings. DCR then executes these rules on meaningful concept truth degrees to provide a final interpretable and semantically-consistent prediction in a differentiable manner. Our experiments show that DCR: (i) improves up to +25% w.r.t. state-of-the-art interpretable concept-based models on challenging benchmarks (ii) discovers meaningful logic rules matching known ground truths even in the absence of concept supervision during training, and (iii), facilitates the generation of counterfactual examples providing the learnt rules as guidance.  \n1. Introduction  \nThe opaque decision process of deep learning (DL) models has failed to inspire human trust despite their state-of-the-art performance across multiple tasks (Rudin, 2019; Bussone  \n*Equal contribution 1University of Cambridge, Cambridge, UK 2Universit Cte d’Azur, Inria, CNRS, I3S, Maasai, Nice, France 3University of Siena, Siena, Italy 4INRA, Universit ParisSaclay, Thiverval-Grignon, France 5 KU Leuven, Leuven, Belgium. Correspondence to: Pietro Barbiero \u003C[pb737@cam.ac.uk](pb737@cam.ac.uk)>.  \nProceedings of the 40 th International Conference on Machine Learning, Honolulu, Hawaii, USA. PMLR 202, 2023 . Copyright 2023 by the author(s) .  \net al., 2015), raising ethical (Durn & Jongsma, 2021; Lo Piano, 2020) and legal (Wachter et al., 2017; EUGDPR, 2017) concerns. For this reason, interpretability is now a core research topic in the field of responsible AI (Rudin, 2019) .  \nConcept-based models (Kim et al., 2018; Chen et al., 2020) aim to increase human trust in deep learning models by using human-understandable concepts to train interpretable models—such as logistic regression or decision trees (Rudin, 2019; Koh et al., 2020; Kazhdan et al., 2020) (Figure 1) . This approach significantly increases human trust in the AI predictor (Rudin, 2019; Shen, 2022) as it allows users to clearly understand a model’s decision process. However, state-of-the-art concept-based models, which rely on concept embeddings (Yeh et al., 2020; Kazhdan et al., 2020; Mahinpei et al., 2021; Espinosa Zarlenga et al., 2022) to attain high performance, are not completely interpretable. Indeed, concept embeddings lack clear semantics on individual dimensions, e.g., ˆcyellow = [2 .3 , 0.3 , −3 .5 ,... ]T does not have semantics assigned to each of its dimensions. This sacrifice of interpretability in favour of model capacity leads to a possible reduction in human trust when using these models, as argued by Rudin (2019); Mahinpei et al. (2021) .  \nIn this paper, we propose the Deep Concept Reasoner 1 (DCR, Section 3), the first interpretable concept-based model building on concept embeddings. DCR applies differentiable and learnable modules on concept embeddingsto build a set of fuzzy rules which can then be executed on semantically meaningful concept truth degrees to provide a final interpretable prediction. Our experiments (Section 4) show that DCR: (i) attains better task accuracy than state-ofthe-art interpretable concept-based mode","cbCaimoXGNKOC8Ho","https://ap.wps.com/l/cbCaimoXGNKOC8Ho","pdf",1890501,1,25,"English","en",105,"# Abstract\n# Introduction\n## Concept-based models and interpretability challenges\n# Preliminaries\n## Concept-based models\n## Concept encoders and supervision signals\n# Deep Concept Reasoner (DCR)\n## Differentiable rule construction from embeddings\n## Executing fuzzy rules on concept truth degrees\n# Experiments\n## Accuracy improvement on benchmarks\n## Discovering logic rules\n## Generating counterfactual examples","[{\"question\":\"Why do concept-based models still face interpretability issues?\",\"answer\":\"They often rely on concept embeddings where individual embedding dimensions have unclear or missing semantic meanings, reducing transparency of the decision process.\"},{\"question\":\"How does the Deep Concept Reasoner (DCR) improve interpretability?\",\"answer\":\"DCR uses neural networks to build differentiable syntactic rule structures from concept embeddings, then executes these rules on semantically meaningful fuzzy concept truth degrees to produce interpretable predictions.\"},{\"question\":\"What evidence supports DCR’s effectiveness?\",\"answer\":\"Experiments report up to +25% improvements over state-of-the-art interpretable concept-based models, discovery of meaningful logic rules that match ground truths, and facilitation of counterfactual example generation using the learned rules.\"}]","Interpretable Neural-Symbolic Concept Reasoning - Deep Concept Reasoner (DCR) | PDF",1785941801,63,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":87,"head_meta":89,"extra_data":91,"updated_unix":28},"interpretable-neural-symbolic-concept-reasoning-deep-concept-reasoner-dcr-127798","",{"@graph":36,"@context":86},[37,54,69],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/interpretable-neural-symbolic-concept-reasoning-deep-concept-reasoner-dcr-127798/127798/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-22","2026-08-05",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"Why do concept-based models still face interpretability issues?","Question",{"text":76,"@type":77},"They often rely on concept embeddings where individual embedding dimensions have unclear or missing semantic meanings, reducing transparency of the decision process.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How does the Deep Concept Reasoner (DCR) improve interpretability?",{"text":81,"@type":77},"DCR uses neural networks to build differentiable syntactic rule structures from concept embeddings, then executes these rules on semantically meaningful fuzzy concept truth degrees to produce interpretable predictions.",{"name":83,"@type":74,"acceptedAnswer":84},"What evidence supports DCR’s effectiveness?",{"text":85,"@type":77},"Experiments report up to +25% improvements over state-of-the-art interpretable concept-based models, discovery of meaningful logic rules that match ground truths, and facilitation of counterfactual example generation using the learned rules.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,124,129,132,136],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":46,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":46,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":46,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":46,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":46,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":46,"category_name":138,"show_sort_weight":107,"slug":139},19,"General","general"]