{"id":"https://openalex.org/W4416432166","doi":"https://doi.org/10.48550/arxiv.2511.00833","title":"Linear Differential Vision Transformer: Learning Visual Contrasts via Pairwise Differentials","display_name":"Linear Differential Vision Transformer: Learning Visual Contrasts via Pairwise Differentials","publication_year":2025,"publication_date":"2025-11-02","ids":{"openalex":"https://openalex.org/W4416432166","doi":"https://doi.org/10.48550/arxiv.2511.00833"},"language":null,"primary_location":{"id":"pmh:oai:arXiv.org:2511.00833","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2511.00833","pdf_url":"https://arxiv.org/pdf/2511.00833","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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"},"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2511.00833","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5035205972","display_name":"Yifan Pu","orcid":"https://orcid.org/0000-0002-0404-1737"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Pu, Yifan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Ying, Jixuan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ying, Jixuan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5107503613","display_name":"Qixiu Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Qixiu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066799604","display_name":"Tianzhu Ye","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ye, Tianzhu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102210006","display_name":"Dongchen Han","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Han, Dongchen","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100401907","display_name":"Xiaochen Wang","orcid":"https://orcid.org/0000-0001-8479-0780"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Xiaochen","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101731051","display_name":"Ziyi Wang","orcid":"https://orcid.org/0000-0002-7752-7439"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Ziyi","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001189952","display_name":"Xinyu Shao","orcid":"https://orcid.org/0000-0001-8080-652X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shao, Xinyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5013240918","display_name":"Gao Huang","orcid":"https://orcid.org/0000-0002-7251-0988"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Gao","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5100754504","display_name":"Xiu Li","orcid":"https://orcid.org/0000-0003-0403-1923"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Xiu","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":10,"corresponding_author_ids":["https://openalex.org/A5035205972"],"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.37790000438690186,"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"}},"topics":[{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.37790000438690186,"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"}},{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.2151000052690506,"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":0.13619999587535858,"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/pooling","display_name":"Pooling","score":0.5950000286102295},{"id":"https://openalex.org/keywords/pairwise-comparison","display_name":"Pairwise comparison","score":0.5710999965667725},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.5562999844551086},{"id":"https://openalex.org/keywords/quadratic-equation","display_name":"Quadratic equation","score":0.5414000153541565},{"id":"https://openalex.org/keywords/differential","display_name":"Differential (mechanical device)","score":0.4814999997615814},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.39590001106262207},{"id":"https://openalex.org/keywords/computational-complexity-theory","display_name":"Computational complexity theory","score":0.3695000112056732},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.35929998755455017},{"id":"https://openalex.org/keywords/image","display_name":"Image (mathematics)","score":0.3555000126361847}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6037999987602234},{"id":"https://openalex.org/C70437156","wikidata":"https://www.wikidata.org/wiki/Q7228652","display_name":"Pooling","level":2,"score":0.5950000286102295},{"id":"https://openalex.org/C184898388","wikidata":"https://www.wikidata.org/wiki/Q1435712","display_name":"Pairwise comparison","level":2,"score":0.5710999965667725},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.5562999844551086},{"id":"https://openalex.org/C129844170","wikidata":"https://www.wikidata.org/wiki/Q41299","display_name":"Quadratic equation","level":2,"score":0.5414000153541565},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5098999738693237},{"id":"https://openalex.org/C93226319","wikidata":"https://www.wikidata.org/wiki/Q193137","display_name":"Differential (mechanical device)","level":2,"score":0.4814999997615814},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.45809999108314514},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.39590001106262207},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.38499999046325684},{"id":"https://openalex.org/C179799912","wikidata":"https://www.wikidata.org/wiki/Q205084","display_name":"Computational complexity theory","level":2,"score":0.3695000112056732},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.36149999499320984},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.35929998755455017},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.3555000126361847},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3483999967575073},{"id":"https://openalex.org/C89600930","wikidata":"https://www.wikidata.org/wiki/Q1423946","display_name":"Segmentation","level":2,"score":0.3012000024318695},{"id":"https://openalex.org/C9417928","wikidata":"https://www.wikidata.org/wiki/Q1070689","display_name":"Image processing","level":3,"score":0.29330000281333923},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.2924000024795532},{"id":"https://openalex.org/C9652623","wikidata":"https://www.wikidata.org/wiki/Q190109","display_name":"Field (mathematics)","level":2,"score":0.2865000069141388},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.27900001406669617},{"id":"https://openalex.org/C190470478","wikidata":"https://www.wikidata.org/wiki/Q2370229","display_name":"Invariant (physics)","level":2,"score":0.27320000529289246},{"id":"https://openalex.org/C205203396","wikidata":"https://www.wikidata.org/wiki/Q612143","display_name":"Bilinear interpolation","level":2,"score":0.27090001106262207},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.2637999951839447},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.2612999975681305},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.2563999891281128},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.25519999861717224},{"id":"https://openalex.org/C124504099","wikidata":"https://www.wikidata.org/wiki/Q56933","display_name":"Image segmentation","level":3,"score":0.25209999084472656}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2511.00833","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2511.00833","pdf_url":"https://arxiv.org/pdf/2511.00833","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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"},{"id":"doi:10.48550/arxiv.2511.00833","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2511.00833","pdf_url":null,"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":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2511.00833","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2511.00833","pdf_url":"https://arxiv.org/pdf/2511.00833","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Vision":[0,218],"Transformers":[1],"(ViTs)":[2],"have":[3],"become":[4],"a":[5,23,47,85,96,121,211],"universal":[6],"backbone":[7],"for":[8,27,50,193],"both":[9,169,202],"image":[10,13],"recognition":[11],"and":[12,99,128,145,172,196,216],"generation.":[14],"Yet":[15],"their":[16],"Multi-Head":[17],"Self-Attention":[18],"(MHSA)":[19],"layer":[20],"still":[21],"performs":[22],"quadratic":[24],"query-key":[25],"interaction":[26,104],"every":[28],"token":[29],"pair,":[30],"spending":[31],"the":[32,61,199,205],"bulk":[33],"of":[34,57,87],"computation":[35],"on":[36,138],"visually":[37],"weak":[38],"or":[39],"redundant":[40],"correlations.":[41],"We":[42],"introduce":[43],"Visual-Contrast":[44],"Attention":[45],"(VCA),":[46],"drop-in":[48],"replacement":[49],"MHSA":[51],"that":[52,179],"injects":[53],"an":[54],"explicit":[55],"notion":[56],"discrimination":[58],"while":[59,155],"reducing":[60],"theoretical":[62],"complexity":[63],"from":[64,111,140],"O(N":[65,69],"N":[66],"C)":[67,71],"to":[68,120,142,153,165],"n":[70,73],"with":[72],"<<":[74],"N.":[75],"VCA":[76,133,208],"first":[77],"distils":[78],"each":[79],"head's":[80],"dense":[81],"query":[82],"field":[83],"into":[84,95],"handful":[86],"spatially":[88],"pooled":[89],"visual-contrast":[90],"tokens,":[91],"then":[92],"splits":[93],"them":[94],"learnable":[97],"positive":[98],"negative":[100],"stream":[101],"whose":[102],"differential":[103],"highlights":[105],"what":[106],"truly":[107],"separates":[108],"one":[109],"region":[110],"another.":[112],"The":[113,220],"module":[114],"adds":[115],"fewer":[116],"than":[117],"0.3M":[118],"parameters":[119],"DeiT-Tiny":[122,135],"backbone,":[123],"requires":[124],"no":[125],"extra":[126],"FLOPs,":[127],"is":[129,223],"wholly":[130],"architecture-agnostic.":[131],"Empirically,":[132],"lifts":[134],"top-1":[136],"accuracy":[137],"ImageNet-1K":[139],"72.2%":[141],"75.6%":[143],"(+3.4)":[144],"improves":[146],"three":[147],"strong":[148],"hierarchical":[149],"ViTs":[150],"by":[151,163],"up":[152],"3.1%,":[154],"in":[156,201],"class-conditional":[157],"ImageNet":[158],"generation":[159],"it":[160],"lowers":[161],"FID-50K":[162],"2.1":[164],"5.2":[166],"points":[167],"across":[168],"diffusion":[170],"(DiT)":[171],"flow":[173],"(SiT)":[174],"models.":[175],"Extensive":[176],"ablations":[177],"confirm":[178],"(i)":[180],"spatial":[181],"pooling":[182],"supplies":[183],"low-variance":[184],"global":[185],"cues,":[186],"(ii)":[187],"dual":[188],"positional":[189],"embeddings":[190],"are":[191],"indispensable":[192],"contrastive":[194],"reasoning,":[195],"(iii)":[197],"combining":[198],"two":[200],"stages":[203],"yields":[204],"strongest":[206],"synergy.":[207],"therefore":[209],"offers":[210],"simple":[212],"path":[213],"towards":[214],"faster":[215],"sharper":[217],"Transformers.":[219],"source":[221],"code":[222],"available":[224],"at":[225],"https://github.com/LeapLabTHU/LinearDiff.":[226]},"counts_by_year":[],"updated_date":"2026-04-17T18:11:37.981687","created_date":"2025-11-06T00:00:00"}