[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-127774-en":3,"doc-seo-127774-105":31,"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":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":28,"seo_description":14,"update_tm":29,"read_time":30},127774,1099523882182,"Eliana","https://ap-avatar.wpscdn.com/davatar_6f874abed73319feea01a86fa6f0fab8",8,"Research & Report","Neuro-Symbolic Continual Learning - Knowledge, Reasoning Shortcuts and Concept Rehearsal","Neuro-Symbolic Continual Learning studies how a model solves a sequence of neuro-symbolic tasks that map subsymbolic inputs to high-level concepts and produce predictions via consistent reasoning with prior knowledge. The work highlights that neuro-symbolic tasks may differ yet share concepts whose semantics stay stable over time. Conventional continual strategies overlook knowledge, while existing neuro-symbolic architectures suffer catastrophic forgetting. Combining continual strategies with neurosymbolic architectures can help forgetting, but may introduce reasoning shortcuts that distort acquired concept semantics and degrade continual performance. The paper proposes COOL, a concept-level continual learning strategy that learns and retains high-quality concepts. Experiments on three novel benchmarks show sustained high performance when other methods fail.","This is a pre print version of the following article:  \nNeuro-Symbolic Continual Learning: Knowledge, Reasoning Shortcuts and Concept Rehearsal / Marconato, E.; Bontempo, G.; Ficarra, E.; Calderara, S.; Passerini, A.; Teso, S. . -202:(2023), pp. 23915-23936.(Intervento presentato al convegno 40th International Conference on Machine Learning, ICML 2023 tenutosi a Honolulu, USA nel 2023) .  \nML Research Press Terms of use:  \nThe terms and conditions for the reuse of this version of the manuscript are specified in the publishing policy. For all terms of use and more information see the publisher's website.  \n01/09/2024 14:57  \n(Article begins on next page)  \nNeuro-Symbolic Continual Learning: Knowledge, Reasoning Shortcuts and Concept Rehearsal  \nEmanuele Marconato * 1 2 Gianpaolo Bontempo * 1 3 Elisa Ficarra 3 Simone Calderara 3 Andrea Passerini 2  \nStefano Teso 4 2  \narXiv :2302 .01242v2 [ cs .LG] 19 Dec 2023  \nAbstract  \nWe introduce Neuro-Symbolic Continual Learning, where a model has to solve a sequence of neuro-symbolic tasks, that is, it has to map subsymbolic inputs to high-level concepts and compute predictions by reasoning consistently with prior knowledge. Our key observation is that neuro-symbolic tasks, although different, often share concepts whose semantics remains stable over time. Traditional approaches fall short: existing continual strategies ignore knowledge altogether, while stock neuro-symbolic architectures suffer from catastrophic forgetting. We show that leveraging prior knowledge by combining neurosymbolic architectures with continual strategies does help avoid catastrophic forgetting, but also that doing so can yield models affected by reasoning shortcuts. These undermine the semantics of the acquired concepts, even when detailed prior knowledge is provided upfront and inference is exact, and in turn continual performance. To overcome these issues, we introduce COOL, a COncept-level cOntinual Learning strategy tailored for neuro-symbolic continual problems that acquires high-quality concepts and remembers them over time. Our experiments on three novel benchmarks highlights how COOL attains sustained high performance on neuro-symbolic continual learning tasks in which other strategies fail.1  \n*Equal contribution 1University of Pisa, Italy 2DISI, University of Trento, Italy 3University of Modena and Reggio Emilia, Italy 4 CIMeC, University of Trento, Italy. Correspondence to: Emanuele Marconato \u003Cemanuele.marconato@unitn.it> .  \nProceedings of the 40 th International Conference on Machine Learning, Honolulu, Hawaii, USA. PMLR 202, 2023 . Copyright 2023 by the author(s) .  \n1The data and code are available at [https://github.com/ema](https://github.com/ema)marconato/NeSy-CL  \n1. Introduction  \nWe initiate the study of Neuro-Symbolic Continual Learning (NeSy-CL), in which the goal is to solve a sequence of neuro-symbolic tasks. As is common in neuro-symbolic (NeSy) prediction (Manhaeve et al., 2018 ; Xu et al., 2018 ; Giunchiglia & Lukasiewicz, 2020 ; Hoernle et al., 2022 ; Ahmed et al., 2022a), the machine is provided prior knowledge relating one or more target labels to symbolic, highlevel concepts extracted from sub-symbolic data, and has to compute a prediction by reasoning over said concepts. The central challenge of Nesy-CL is that the data distribution and the knowledge may vary across tasks. E.g., in medical diagnosis knowledge may encode known relationships between possible symptoms and conditions, while different tasks are characterized by different distributions of X-rayscans, symptoms and conditions. The goal, as in continual learning (CL) (Parisi et al., 2019), is to obtain a model that attains high accuracy on new tasks without forgetting what it has already learned under a limited storage budget.  \nExisting approaches are insufficient for NeSy-CL: neurosymbolic models are designed for offline learning and as such suffer from catastrophic forgetting (Parisi et al., 2019), while continual learn","cbCaibWRVF9aYqq1","https://ap.wps.com/l/cbCaibWRVF9aYqq1","pdf",7642677,2,1,23,"English","en",105,"# Abstract\n# Introduction\n## Neuro-symbolic tasks and prior knowledge\n## Central challenge: distribution and knowledge shift\n## Limitations of existing approaches","[{\"question\":\"What is the goal of Neuro-Symbolic Continual Learning?\",\"answer\":\"It aims to solve a sequence of neuro-symbolic tasks, mapping subsymbolic inputs to high-level concepts and reasoning consistently with prior knowledge, while handling changes across tasks.\"},{\"question\":\"Why do existing approaches fail in this setting?\",\"answer\":\"Continual learning strategies often ignore prior knowledge, and standard neuro-symbolic architectures can suffer catastrophic forgetting. Even when combining the two, models may acquire reasoning shortcuts that distort concept semantics and hurt continual performance.\"},{\"question\":\"What does COOL do to address reasoning shortcuts and forgetting?\",\"answer\":\"COOL is a concept-level continual learning strategy that acquires high-quality concepts and remembers them over time, improving sustained performance on neuro-symbolic continual learning benchmarks.\"}]","Neuro-Symbolic Continual Learning - Knowledge, Reasoning Shortcuts and Concept Rehearsal | PDF",1785941534,58,{"code":4,"msg":32,"data":33},"ok",{"site_id":25,"language":24,"slug":34,"title":13,"keywords":35,"description":14,"schema_data":36,"social_meta":87,"head_meta":89,"extra_data":91,"updated_unix":29},"neuro-symbolic-continual-learning-knowledge-reasoning-shortcuts-and-concept-rehearsal","",{"@graph":37,"@context":86},[38,54,69],{"@type":39,"itemListElement":40},"BreadcrumbList",[41,45,48,51],{"item":42,"name":43,"@type":44,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":46,"name":47,"@type":44,"position":20},"https://docshare.wps.com/document/","Document",{"item":49,"name":12,"@type":44,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":44,"position":53},"https://docshare.wps.com/document/neuro-symbolic-continual-learning-knowledge-reasoning-shortcuts-and-concept-rehearsal/127774/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":24,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":42,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-23","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},"What is the goal of Neuro-Symbolic Continual Learning?","Question",{"text":76,"@type":77},"It aims to solve a sequence of neuro-symbolic tasks, mapping subsymbolic inputs to high-level concepts and reasoning consistently with prior knowledge, while handling changes across tasks.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"Why do existing approaches fail in this setting?",{"text":81,"@type":77},"Continual learning strategies often ignore prior knowledge, and standard neuro-symbolic architectures can suffer catastrophic forgetting. Even when combining the two, models may acquire reasoning shortcuts that distort concept semantics and hurt continual performance.",{"name":83,"@type":74,"acceptedAnswer":84},"What does COOL do to address reasoning shortcuts and forgetting?",{"text":85,"@type":77},"COOL is a concept-level continual learning strategy that acquires high-quality concepts and remembers them over time, improving sustained performance on neuro-symbolic continual learning benchmarks.","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":25},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,124,129,132,136],{"id":21,"doc_module":4,"doc_module_name":47,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":20,"doc_module":4,"doc_module_name":47,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":47,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":47,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":47,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":47,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":47,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":47,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":47,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":47,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":47,"category_name":138,"show_sort_weight":107,"slug":139},19,"General","general"]