{"id":"https://openalex.org/W7138377663","doi":"https://doi.org/10.1609/aaai.v40i24.39046","title":"Mechanistic Dissection of Cross-Attention Subspaces in Text-to-Image Diffusion Models","display_name":"Mechanistic Dissection of Cross-Attention Subspaces in Text-to-Image Diffusion Models","publication_year":2026,"publication_date":"2026-03-14","ids":{"openalex":"https://openalex.org/W7138377663","doi":"https://doi.org/10.1609/aaai.v40i24.39046"},"language":"en","primary_location":{"id":"doi:10.1609/aaai.v40i24.39046","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i24.39046","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://doi.org/10.1609/aaai.v40i24.39046","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5058398283","display_name":"Jun-Hyun Bae","orcid":"https://orcid.org/0009-0004-0593-5670"},"institutions":[{"id":"https://openalex.org/I31419693","display_name":"Kyungpook National University","ror":"https://ror.org/040c17130","country_code":"KR","type":"education","lineage":["https://openalex.org/I31419693"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jun-Hyun Bae","raw_affiliation_strings":["Kyungpook National University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Kyungpook National University","institution_ids":["https://openalex.org/I31419693"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129654574","display_name":"Wonyong Jo","orcid":null},"institutions":[{"id":"https://openalex.org/I31419693","display_name":"Kyungpook National University","ror":"https://ror.org/040c17130","country_code":"KR","type":"education","lineage":["https://openalex.org/I31419693"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Wonyong Jo","raw_affiliation_strings":["Kyungpook National University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Kyungpook National University","institution_ids":["https://openalex.org/I31419693"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5002247218","display_name":"Jaehyup Lee","orcid":"https://orcid.org/0000-0002-2137-2650"},"institutions":[{"id":"https://openalex.org/I31419693","display_name":"Kyungpook National University","ror":"https://ror.org/040c17130","country_code":"KR","type":"education","lineage":["https://openalex.org/I31419693"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jaehyup Lee","raw_affiliation_strings":["Kyungpook National University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Kyungpook National University","institution_ids":["https://openalex.org/I31419693"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5128657244","display_name":"Heechul Jung","orcid":null},"institutions":[{"id":"https://openalex.org/I31419693","display_name":"Kyungpook National University","ror":"https://ror.org/040c17130","country_code":"KR","type":"education","lineage":["https://openalex.org/I31419693"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Heechul Jung","raw_affiliation_strings":["Kyungpook National University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Kyungpook National University","institution_ids":["https://openalex.org/I31419693"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.37623762,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"40","issue":"24","first_page":"19657","last_page":"19665"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11094","display_name":"Face Recognition and Perception","score":0.31189998984336853,"subfield":{"id":"https://openalex.org/subfields/2805","display_name":"Cognitive Neuroscience"},"field":{"id":"https://openalex.org/fields/28","display_name":"Neuroscience"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},"topics":[{"id":"https://openalex.org/T11094","display_name":"Face Recognition and Perception","score":0.31189998984336853,"subfield":{"id":"https://openalex.org/subfields/2805","display_name":"Cognitive Neuroscience"},"field":{"id":"https://openalex.org/fields/28","display_name":"Neuroscience"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.1696999967098236,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.09369999915361404,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/linear-subspace","display_name":"Linear subspace","score":0.8360000252723694},{"id":"https://openalex.org/keywords/embedding","display_name":"Embedding","score":0.6431999802589417},{"id":"https://openalex.org/keywords/singular-value-decomposition","display_name":"Singular value decomposition","score":0.46860000491142273},{"id":"https://openalex.org/keywords/diffusion","display_name":"Diffusion","score":0.4300000071525574},{"id":"https://openalex.org/keywords/transformation","display_name":"Transformation (genetics)","score":0.4226999878883362},{"id":"https://openalex.org/keywords/term","display_name":"Term (time)","score":0.36469998955726624},{"id":"https://openalex.org/keywords/spectral-analysis","display_name":"Spectral analysis","score":0.3416000008583069},{"id":"https://openalex.org/keywords/encode","display_name":"ENCODE","score":0.3402999937534332}],"concepts":[{"id":"https://openalex.org/C12362212","wikidata":"https://www.wikidata.org/wiki/Q728435","display_name":"Linear subspace","level":2,"score":0.8360000252723694},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.6431999802589417},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5641000270843506},{"id":"https://openalex.org/C22789450","wikidata":"https://www.wikidata.org/wiki/Q420904","display_name":"Singular value decomposition","level":2,"score":0.46860000491142273},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.4417000114917755},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.4300000071525574},{"id":"https://openalex.org/C204241405","wikidata":"https://www.wikidata.org/wiki/Q461499","display_name":"Transformation (genetics)","level":3,"score":0.4226999878883362},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3767000138759613},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.36469998955726624},{"id":"https://openalex.org/C2983668108","wikidata":"https://www.wikidata.org/wiki/Q280453","display_name":"Spectral analysis","level":3,"score":0.3416000008583069},{"id":"https://openalex.org/C66746571","wikidata":"https://www.wikidata.org/wiki/Q1134833","display_name":"ENCODE","level":3,"score":0.3402999937534332},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.31859999895095825},{"id":"https://openalex.org/C32946077","wikidata":"https://www.wikidata.org/wiki/Q618079","display_name":"Network analysis","level":2,"score":0.30219998955726624},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.30219998955726624},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.29899999499320984},{"id":"https://openalex.org/C18555067","wikidata":"https://www.wikidata.org/wiki/Q8375051","display_name":"Joint (building)","level":2,"score":0.28619998693466187},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.2800999879837036},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.2757999897003174},{"id":"https://openalex.org/C186060115","wikidata":"https://www.wikidata.org/wiki/Q30336093","display_name":"Biological system","level":1,"score":0.26409998536109924},{"id":"https://openalex.org/C48209547","wikidata":"https://www.wikidata.org/wiki/Q1331104","display_name":"Controllability","level":2,"score":0.26159998774528503},{"id":"https://openalex.org/C521332185","wikidata":"https://www.wikidata.org/wiki/Q185816","display_name":"Analogy","level":2,"score":0.2574000060558319},{"id":"https://openalex.org/C170133592","wikidata":"https://www.wikidata.org/wiki/Q1806883","display_name":"Latent semantic analysis","level":2,"score":0.257099986076355},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.25270000100135803},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.25220000743865967}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1609/aaai.v40i24.39046","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i24.39046","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},{"id":"pmh:oai:ojs.aaai.org:article/39046","is_oa":false,"landing_page_url":"https://ojs.aaai.org/index.php/AAAI/article/view/39046","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"2159-5399","raw_type":"info:eu-repo/semantics/publishedVersion"}],"best_oa_location":{"id":"doi:10.1609/aaai.v40i24.39046","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i24.39046","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Text-to-image":[0],"diffusion":[1,77],"models":[2],"utilize":[3],"cross-attention":[4,37,65,138],"to":[5,20,128],"integrate":[6],"textual":[7],"information":[8],"into":[9],"the":[10,15,32,76],"visual":[11,96],"latent":[12,21],"space,":[13],"yet":[14],"transformation":[16],"from":[17],"text":[18,105],"embeddings":[19],"features":[22],"remains":[23],"largely":[24],"unexplored.":[25],"We":[26,89],"provide":[27],"a":[28],"mechanistic":[29,110,150],"analysis":[30,41,47],"of":[31,98,146],"output-value":[33],"(OV)":[34],"circuits":[35,63],"within":[36],"layers":[38,139],"through":[39],"spectral":[40,84,118,144],"via":[42],"singular":[43,59],"value":[44],"decomposition.":[45],"Our":[46,134],"reveals":[48,136],"that":[49,80,114],"semantic":[50,141],"concepts":[51,142],"are":[52],"encoded":[53],"in":[54,61,75,143],"low-dimensional":[55],"subspaces":[56,100,145],"spanned":[57],"by":[58,94],"vectors":[60],"OV":[62,147],"across":[64],"heads.":[66],"To":[67],"verify":[68],"this,":[69],"we":[70,112],"intervene":[71],"on":[72,82],"concept-related":[73],"components":[74,85,119],"process,":[78],"demonstrating":[79],"intervention":[81],"identified":[83],"affects":[86],"conceptual":[87],"changes.":[88],"further":[90],"validate":[91],"these":[92,117],"findings":[93],"examining":[95],"outputs":[97],"isolated":[99],"and":[101,152],"their":[102],"alignment":[103],"with":[104,125],"embedding":[106],"space.":[107],"Through":[108],"this":[109],"understanding,":[111],"demonstrate":[113],"only":[115],"nullifying":[116],"can":[120],"achieve":[121],"targeted":[122],"concept":[123,155],"removal":[124],"performance":[126],"comparable":[127],"existing":[129],"methods":[130],"while":[131],"providing":[132,149],"interpretability.":[133],"work":[135],"how":[137],"encode":[140],"circuits,":[148],"insights":[151],"enabling":[153],"precise":[154],"manipulation":[156],"without":[157],"retraining.":[158]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-03-18T00:00:00"}