{"id":"https://openalex.org/W7167852099","doi":"https://doi.org/10.48550/arxiv.2607.07135","title":"Sparse Attention for Dense Open-Vocabulary Prediction in CLIP","display_name":"Sparse Attention for Dense Open-Vocabulary Prediction in CLIP","publication_year":2026,"publication_date":"2026-07-08","ids":{"openalex":"https://openalex.org/W7167852099","doi":"https://doi.org/10.48550/arxiv.2607.07135"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2607.07135","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.07135","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2607.07135","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5093701194","display_name":"Fatimah Zohra","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zohra, Fatimah","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5140335281","display_name":"Chen Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103870133","display_name":"Liu S","orcid":"https://orcid.org/0000-0001-7255-0982"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Shuming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5140342791","display_name":"Bernard Ghanem","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ghanem, Bernard","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.37950000166893005,"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"}},"topics":[{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.37950000166893005,"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/T11605","display_name":"Visual Attention and Saliency Detection","score":0.2062000036239624,"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":0.14030000567436218,"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/softmax-function","display_name":"Softmax function","score":0.823199987411499},{"id":"https://openalex.org/keywords/pascal","display_name":"Pascal (unit)","score":0.6290000081062317},{"id":"https://openalex.org/keywords/segmentation","display_name":"Segmentation","score":0.5455999970436096},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.4961000084877014},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.4474000036716461},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.40720000863075256}],"concepts":[{"id":"https://openalex.org/C188441871","wikidata":"https://www.wikidata.org/wiki/Q7554146","display_name":"Softmax function","level":3,"score":0.823199987411499},{"id":"https://openalex.org/C75608658","wikidata":"https://www.wikidata.org/wiki/Q44395","display_name":"Pascal (unit)","level":2,"score":0.6290000081062317},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6166999936103821},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5985000133514404},{"id":"https://openalex.org/C89600930","wikidata":"https://www.wikidata.org/wiki/Q1423946","display_name":"Segmentation","level":2,"score":0.5455999970436096},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.4961000084877014},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.4474000036716461},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.40720000863075256},{"id":"https://openalex.org/C149441793","wikidata":"https://www.wikidata.org/wiki/Q200726","display_name":"Probability distribution","level":2,"score":0.36730000376701355},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.35089999437332153},{"id":"https://openalex.org/C2986089797","wikidata":"https://www.wikidata.org/wiki/Q6501338","display_name":"Visual attention","level":3,"score":0.33070001006126404},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3237000107765198},{"id":"https://openalex.org/C2778220771","wikidata":"https://www.wikidata.org/wiki/Q1522579","display_name":"Substitution (logic)","level":2,"score":0.323199987411499},{"id":"https://openalex.org/C197096303","wikidata":"https://www.wikidata.org/wiki/Q869887","display_name":"Probability mass function","level":3,"score":0.30410000681877136},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2863999903202057},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2777000069618225},{"id":"https://openalex.org/C2984485829","wikidata":"https://www.wikidata.org/wiki/Q100766421","display_name":"Noise level","level":3,"score":0.27129998803138733}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2607.07135","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.07135","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2607.07135","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.07135","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","score":0.808245837688446,"display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Contrastive":[0],"Language-Image":[1],"Pre-training":[2],"(CLIP)":[3],"relies":[4],"on":[5,120],"softmax-based":[6],"self-attention,":[7],"a":[8,90],"strictly":[9],"positive":[10],"distribution":[11],"that":[12,46,93],"assigns":[13],"probability":[14],"mass":[15,112],"to":[16,98,142],"every":[17],"pair":[18],"of":[19,63],"tokens-even":[20],"semantically":[21],"irrelevant":[22,108],"ones.":[23],"While":[24],"these":[25],"dense":[26],"softmax":[27,66],"weights":[28],"are":[29],"effective":[30],"for":[31,54],"gathering":[32],"broad":[33],"context":[34],"during":[35],"pre-training,":[36],"they":[37],"spread":[38],"attention":[39,83,138,147],"across":[40,78],"many":[41],"low-salience":[42],"tokens,":[43],"producing":[44],"noise":[45],"obscures":[47],"the":[48,64,68,74,80,114,135,145,150],"fine-grained,":[49],"spatially":[50],"localized":[51],"cues":[52],"required":[53],"dense,":[55],"open-vocabulary":[56,121],"prediction.":[57],"We":[58,118],"study":[59],"an":[60,103],"inference-time":[61],"substitution":[62],"row-wise":[65],"in":[67],"final":[69],"visual":[70],"self-attention":[71],"layers":[72],"with":[73],"$\u03b1$-entmax":[75],"transform,":[76],"applied":[77],"both":[79],"standard":[81],"query-key":[82],"and":[84,130],"self-correlation":[85],"variants.":[86],"Because":[87],"entmax":[88],"applies":[89],"data-dependent":[91],"threshold":[92],"maps":[94],"low":[95],"scores":[96],"exactly":[97],"zero,":[99],"it":[100],"acts":[101],"as":[102],"implicit":[104],"denoiser,":[105],"zeroing":[106],"contextually":[107],"dependencies":[109],"while":[110],"redistributing":[111],"onto":[113],"most":[115],"relevant":[116],"tokens.":[117],"evaluate":[119],"tasks-dense":[122],"semantic":[123],"segmentation":[124],"(Pascal":[125],"VOC,":[126],"Pascal":[127],"Context,":[128],"ADE20K)":[129],"fine-grained":[131],"retrieval":[132],"(FG-OVD)-and":[133],"find":[134],"gain":[136],"from":[137],"sparsification":[139],"is":[140],"proportional":[141],"how":[143],"much":[144],"baseline":[146],"spreads":[148],"off":[149],"target":[151],"class.":[152]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-07-10T00:00:00"}