{"id":"https://openalex.org/W4414076984","doi":"https://doi.org/10.1017/s1471068425100318","title":"Symbolic Rule Extraction From Attention-Guided Sparse Representations in Vision Transformers","display_name":"Symbolic Rule Extraction From Attention-Guided Sparse Representations in Vision Transformers","publication_year":2025,"publication_date":"2025-07-01","ids":{"openalex":"https://openalex.org/W4414076984","doi":"https://doi.org/10.1017/s1471068425100318"},"language":"en","primary_location":{"id":"doi:10.1017/s1471068425100318","is_oa":true,"landing_page_url":"https://doi.org/10.1017/s1471068425100318","pdf_url":"https://www.cambridge.org/core/services/aop-cambridge-core/content/view/9F56FD1261B3A4D12CBE57E9DBF6A4A3/S1471068425100318a.pdf/div-class-title-symbolic-rule-extraction-from-attention-guided-sparse-representations-in-vision-transformers-div.pdf","source":{"id":"https://openalex.org/S59670734","display_name":"Theory and Practice of Logic Programming","issn_l":"1471-0684","issn":["1471-0684","1475-3081"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310311721","host_organization_name":"Cambridge University Press","host_organization_lineage":["https://openalex.org/P4310311721","https://openalex.org/P4310311702"],"host_organization_lineage_names":["Cambridge University Press","University of Cambridge"],"type":"journal"},"license":"cc-by-nd","license_id":"https://openalex.org/licenses/cc-by-nd","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Theory and Practice of Logic Programming","raw_type":"journal-article"},"type":"article","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://www.cambridge.org/core/services/aop-cambridge-core/content/view/9F56FD1261B3A4D12CBE57E9DBF6A4A3/S1471068425100318a.pdf/div-class-title-symbolic-rule-extraction-from-attention-guided-sparse-representations-in-vision-transformers-div.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5058520906","display_name":"Parth Padalkar","orcid":"https://orcid.org/0000-0003-1015-0777"},"institutions":[{"id":"https://openalex.org/I162577319","display_name":"The University of Texas at Dallas","ror":"https://ror.org/049emcs32","country_code":"US","type":"education","lineage":["https://openalex.org/I162577319"]},{"id":"https://openalex.org/I87573096","display_name":"University of North Texas at Dallas","ror":"https://ror.org/01n51v443","country_code":"US","type":"education","lineage":["https://openalex.org/I87573096"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"PARTH PADALKAR","raw_affiliation_strings":["The University of Texas at Dallas","The University of Texas at Dallas, USA (e-mails:"],"raw_orcid":"https://orcid.org/0000-0003-1015-0777","affiliations":[{"raw_affiliation_string":"The University of Texas at Dallas","institution_ids":["https://openalex.org/I162577319"]},{"raw_affiliation_string":"The University of Texas at Dallas, USA (e-mails:","institution_ids":["https://openalex.org/I162577319","https://openalex.org/I87573096"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5067377863","display_name":"Gopal Gupta","orcid":"https://orcid.org/0000-0001-9727-0362"},"institutions":[{"id":"https://openalex.org/I162577319","display_name":"The University of Texas at Dallas","ror":"https://ror.org/049emcs32","country_code":"US","type":"education","lineage":["https://openalex.org/I162577319"]},{"id":"https://openalex.org/I87573096","display_name":"University of North Texas at Dallas","ror":"https://ror.org/01n51v443","country_code":"US","type":"education","lineage":["https://openalex.org/I87573096"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"GOPAL GUPTA","raw_affiliation_strings":["The University of Texas at Dallas","The University of Texas at Dallas, USA (e-mails:"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Texas at Dallas","institution_ids":["https://openalex.org/I162577319"]},{"raw_affiliation_string":"The University of Texas at Dallas, USA (e-mails:","institution_ids":["https://openalex.org/I162577319","https://openalex.org/I87573096"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.6534,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":{"value":0.67744591,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":97,"max":98},"biblio":{"volume":"25","issue":"4","first_page":"722","last_page":"738"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11992","display_name":"CCD and CMOS Imaging Sensors","score":0.9958000183105469,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11992","display_name":"CCD and CMOS Imaging Sensors","score":0.9958000183105469,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11605","display_name":"Visual Attention and Saliency Detection","score":0.9939000010490417,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9908000230789185,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/executable","display_name":"Executable","score":0.5723999738693237},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.5608999729156494},{"id":"https://openalex.org/keywords/modular-design","display_name":"Modular design","score":0.4993000030517578},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.43130001425743103},{"id":"https://openalex.org/keywords/convolutional-neural-network","display_name":"Convolutional neural network","score":0.4226999878883362},{"id":"https://openalex.org/keywords/preprocessor","display_name":"Preprocessor","score":0.40139999985694885},{"id":"https://openalex.org/keywords/sparse-approximation","display_name":"Sparse approximation","score":0.3986000120639801},{"id":"https://openalex.org/keywords/classifier","display_name":"Classifier (UML)","score":0.397599995136261},{"id":"https://openalex.org/keywords/symbolic-data-analysis","display_name":"Symbolic data analysis","score":0.36880001425743103}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8672999739646912},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6111999750137329},{"id":"https://openalex.org/C160145156","wikidata":"https://www.wikidata.org/wiki/Q778586","display_name":"Executable","level":2,"score":0.5723999738693237},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5608999729156494},{"id":"https://openalex.org/C101468663","wikidata":"https://www.wikidata.org/wiki/Q1620158","display_name":"Modular design","level":2,"score":0.4993000030517578},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.43130001425743103},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.4226999878883362},{"id":"https://openalex.org/C34736171","wikidata":"https://www.wikidata.org/wiki/Q918333","display_name":"Preprocessor","level":2,"score":0.40139999985694885},{"id":"https://openalex.org/C124066611","wikidata":"https://www.wikidata.org/wiki/Q28684319","display_name":"Sparse approximation","level":2,"score":0.3986000120639801},{"id":"https://openalex.org/C95623464","wikidata":"https://www.wikidata.org/wiki/Q1096149","display_name":"Classifier (UML)","level":2,"score":0.397599995136261},{"id":"https://openalex.org/C65620979","wikidata":"https://www.wikidata.org/wiki/Q7661176","display_name":"Symbolic data analysis","level":2,"score":0.36880001425743103},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.3375999927520752},{"id":"https://openalex.org/C58166","wikidata":"https://www.wikidata.org/wiki/Q224821","display_name":"Fuzzy logic","level":2,"score":0.33719998598098755},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.33570000529289246},{"id":"https://openalex.org/C2781067378","wikidata":"https://www.wikidata.org/wiki/Q17027399","display_name":"Interpretability","level":2,"score":0.3239000141620636},{"id":"https://openalex.org/C77637269","wikidata":"https://www.wikidata.org/wiki/Q7002051","display_name":"Neural coding","level":2,"score":0.3124000132083893},{"id":"https://openalex.org/C106301342","wikidata":"https://www.wikidata.org/wiki/Q4117933","display_name":"Entropy (arrow of time)","level":2,"score":0.301800012588501},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.3001999855041504},{"id":"https://openalex.org/C52622490","wikidata":"https://www.wikidata.org/wiki/Q1026626","display_name":"Feature extraction","level":2,"score":0.295199990272522},{"id":"https://openalex.org/C85847156","wikidata":"https://www.wikidata.org/wiki/Q59015987","display_name":"Verifiable secret sharing","level":3,"score":0.29490000009536743},{"id":"https://openalex.org/C161301231","wikidata":"https://www.wikidata.org/wiki/Q3478658","display_name":"Knowledge representation and reasoning","level":2,"score":0.28850001096725464},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.28519999980926514},{"id":"https://openalex.org/C2779382394","wikidata":"https://www.wikidata.org/wiki/Q1464197","display_name":"Inductive logic programming","level":2,"score":0.2851000130176544},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.2651999890804291},{"id":"https://openalex.org/C2776095079","wikidata":"https://www.wikidata.org/wiki/Q489538","display_name":"The Symbolic","level":2,"score":0.26190000772476196},{"id":"https://openalex.org/C102993220","wikidata":"https://www.wikidata.org/wiki/Q387196","display_name":"Description logic","level":2,"score":0.25929999351501465},{"id":"https://openalex.org/C197352929","wikidata":"https://www.wikidata.org/wiki/Q1074074","display_name":"Inductive bias","level":4,"score":0.25920000672340393},{"id":"https://openalex.org/C159423971","wikidata":"https://www.wikidata.org/wiki/Q177251","display_name":"Associative property","level":2,"score":0.25690001249313354},{"id":"https://openalex.org/C36299963","wikidata":"https://www.wikidata.org/wiki/Q1369844","display_name":"Observability","level":2,"score":0.2540999948978424}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1017/s1471068425100318","is_oa":true,"landing_page_url":"https://doi.org/10.1017/s1471068425100318","pdf_url":"https://www.cambridge.org/core/services/aop-cambridge-core/content/view/9F56FD1261B3A4D12CBE57E9DBF6A4A3/S1471068425100318a.pdf/div-class-title-symbolic-rule-extraction-from-attention-guided-sparse-representations-in-vision-transformers-div.pdf","source":{"id":"https://openalex.org/S59670734","display_name":"Theory and Practice of Logic Programming","issn_l":"1471-0684","issn":["1471-0684","1475-3081"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310311721","host_organization_name":"Cambridge University Press","host_organization_lineage":["https://openalex.org/P4310311721","https://openalex.org/P4310311702"],"host_organization_lineage_names":["Cambridge University Press","University of Cambridge"],"type":"journal"},"license":"cc-by-nd","license_id":"https://openalex.org/licenses/cc-by-nd","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Theory and Practice of Logic Programming","raw_type":"journal-article"},{"id":"pmh:oai:arXiv.org:2505.06745","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.06745","pdf_url":"https://arxiv.org/pdf/2505.06745","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"doi:10.1017/s1471068425100318","is_oa":true,"landing_page_url":"https://doi.org/10.1017/s1471068425100318","pdf_url":"https://www.cambridge.org/core/services/aop-cambridge-core/content/view/9F56FD1261B3A4D12CBE57E9DBF6A4A3/S1471068425100318a.pdf/div-class-title-symbolic-rule-extraction-from-attention-guided-sparse-representations-in-vision-transformers-div.pdf","source":{"id":"https://openalex.org/S59670734","display_name":"Theory and Practice of Logic Programming","issn_l":"1471-0684","issn":["1471-0684","1475-3081"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310311721","host_organization_name":"Cambridge University Press","host_organization_lineage":["https://openalex.org/P4310311721","https://openalex.org/P4310311702"],"host_organization_lineage_names":["Cambridge University Press","University of Cambridge"],"type":"journal"},"license":"cc-by-nd","license_id":"https://openalex.org/licenses/cc-by-nd","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Theory and Practice of Logic Programming","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G7024220182","display_name":"RI: SMALL: Inducing Answer Set Programs to Provide Accurate and Concise Explanation of Machine-learned Models","funder_award_id":"1910131","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320327708","display_name":"University of Texas at Dallas","ror":"https://ror.org/049emcs32"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4414076984.pdf","grobid_xml":"https://content.openalex.org/works/W4414076984.grobid-xml"},"referenced_works_count":16,"referenced_works":["https://openalex.org/W1606891084","https://openalex.org/W2067713319","https://openalex.org/W2161942583","https://openalex.org/W2311857205","https://openalex.org/W2732026016","https://openalex.org/W2963263783","https://openalex.org/W2997560917","https://openalex.org/W3138516171","https://openalex.org/W3155306957","https://openalex.org/W3161744644","https://openalex.org/W4312443924","https://openalex.org/W4390729846","https://openalex.org/W4391489324","https://openalex.org/W4393148357","https://openalex.org/W4406400595","https://openalex.org/W4408503764"],"related_works":["https://openalex.org/W2772917594","https://openalex.org/W2036807459","https://openalex.org/W2058170566","https://openalex.org/W2755342338","https://openalex.org/W2166024367","https://openalex.org/W3116076068","https://openalex.org/W2229312674","https://openalex.org/W2951359407","https://openalex.org/W2079911747","https://openalex.org/W1969923398"],"abstract_inverted_index":{"Abstract":[0],"Recentneuro-symbolic":[1],"approaches":[2],"have":[3],"successfully":[4],"extracted":[5,140],"symbolic":[6,45,136,183],"rule-sets":[7],"from":[8,48,179],"Convolutional":[9],"Neural":[10],"Network-based":[11],"models":[12],"to":[13,20,27,108,174],"enhance":[14],"interpretability.":[15],"However,":[16],"applying":[17],"similar":[18],"techniques":[19],"Vision":[21],"Transformers":[22],"(ViTs)":[23],"remains":[24],"challenging":[25],"due":[26],"their":[28],"lack":[29],"of":[30,91,119],"modular":[31],"concept":[32,54,102,159],"detectors":[33],"and":[34,69,96,166,191],"reliance":[35],"on":[36,65,156],"global":[37],"self-attention":[38],"mechanisms.":[39],"We":[40],"propose":[41],"a":[42,52,71,89,114,120,126,149,186],"framework":[43],"for":[44,80],"rule":[46],"extraction":[47],"ViTs":[49,180],"by":[50,57],"introducing":[51],"sparse":[53,158,182],"layer":[55,63,152],"inspired":[56],"Sparse":[58],"Autoencoders":[59],"(SAEs).":[60],"This":[61,169],"linear":[62],"operates":[64,154],"attention-weighted":[66],"patch":[67],"representations":[68],"learns":[70],"disentangled,":[72],"binarized":[73,101],"representation":[74],"in":[75,116,189],"which":[76,112],"individual":[77],"neurons":[78],"activate":[79],"high-level":[81],"visual":[82],"concepts.":[83],"To":[84],"encourage":[85],"interpretability,":[86],"we":[87],"apply":[88],"combination":[90],"L1":[92],"sparsity,":[93],"entropy":[94],"minimization,":[95],"supervised":[97],"contrastive":[98],"loss.":[99],"These":[100],"activations":[103],"are":[104,164],"used":[105],"as":[106,148],"input":[107],"the":[109,117,131,139,157,172],"FOLD-SE-M":[110],"algorithm,":[111],"generates":[113],"rule-set":[115,141],"form":[118],"logic":[121,177],"program.":[122],"Our":[123],"method":[124],"achieves":[125],"better":[127],"classification":[128],"accuracy":[129],"than":[130],"standard":[132],"ViT":[133],"while":[134],"enabling":[135],"reasoning.":[137],"Crucially,":[138],"is":[142,171],"not":[143],"merely":[144],"post-hoc":[145],"but":[146],"acts":[147],"logic-based":[150],"decision":[151],"that":[153],"directly":[155],"representations.":[160],"The":[161],"resulting":[162],"programs":[163,178],"concise":[165],"semantically":[167],"meaningful.":[168],"work":[170],"first":[173],"extract":[175],"executable":[176],"using":[181],"representations,":[184],"providing":[185],"step":[187],"forward":[188],"interpretable":[190],"verifiable":[192],"neuro-symbolic":[193],"AI.":[194]},"counts_by_year":[{"year":2026,"cited_by_count":2}],"updated_date":"2026-07-15T18:14:33.161393","created_date":"2025-10-10T00:00:00"}