{"id":"https://openalex.org/W7160898446","doi":"https://doi.org/10.48550/arxiv.2605.10780","title":"Beyond the Last Layer: Multi-Layer Representation Fusion for Visual Tokenization","display_name":"Beyond the Last Layer: Multi-Layer Representation Fusion for Visual Tokenization","publication_year":2026,"publication_date":"2026-05-11","ids":{"openalex":"https://openalex.org/W7160898446","doi":"https://doi.org/10.48550/arxiv.2605.10780"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.10780","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10780","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":"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.2605.10780","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5135983072","display_name":"Xuanyu Zhu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhu, Xuanyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135976942","display_name":"Yan Bai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bai, Yan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135990567","display_name":"Yang Shi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shi, Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5063901345","display_name":"Yihang Lou","orcid":"https://orcid.org/0000-0002-8143-389X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lou, Yihang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135968264","display_name":"Yuanxing Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yuanxing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135967685","display_name":"Jing Jin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jin, Jing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135967361","display_name":"Yuan Zhou","orcid":"https://orcid.org/0000-0001-5685-066X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Yuan","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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.8180999755859375,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.8180999755859375,"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/T11094","display_name":"Face Recognition and Perception","score":0.03440000116825104,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.02590000070631504,"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/encoder","display_name":"Encoder","score":0.7523000240325928},{"id":"https://openalex.org/keywords/representation","display_name":"Representation (politics)","score":0.6920999884605408},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.5640000104904175},{"id":"https://openalex.org/keywords/encoding","display_name":"Encoding (memory)","score":0.5358999967575073},{"id":"https://openalex.org/keywords/constraint","display_name":"Constraint (computer-aided design)","score":0.5059999823570251},{"id":"https://openalex.org/keywords/fusion","display_name":"Fusion","score":0.4424000084400177},{"id":"https://openalex.org/keywords/dimension","display_name":"Dimension (graph theory)","score":0.42149999737739563},{"id":"https://openalex.org/keywords/feature-learning","display_name":"Feature learning","score":0.4198000133037567},{"id":"https://openalex.org/keywords/visualization","display_name":"Visualization","score":0.4162999987602234}],"concepts":[{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.7523000240325928},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7520999908447266},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.6920999884605408},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6312000155448914},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5640000104904175},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.5358999967575073},{"id":"https://openalex.org/C2776036281","wikidata":"https://www.wikidata.org/wiki/Q48769818","display_name":"Constraint (computer-aided design)","level":2,"score":0.5059999823570251},{"id":"https://openalex.org/C158525013","wikidata":"https://www.wikidata.org/wiki/Q2593739","display_name":"Fusion","level":2,"score":0.4424000084400177},{"id":"https://openalex.org/C33676613","wikidata":"https://www.wikidata.org/wiki/Q13415176","display_name":"Dimension (graph theory)","level":2,"score":0.42149999737739563},{"id":"https://openalex.org/C59404180","wikidata":"https://www.wikidata.org/wiki/Q17013334","display_name":"Feature learning","level":2,"score":0.4198000133037567},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.4162999987602234},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.41519999504089355},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.4009999930858612},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.39010000228881836},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.3824000060558319},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.373199999332428},{"id":"https://openalex.org/C101738243","wikidata":"https://www.wikidata.org/wiki/Q786435","display_name":"Autoencoder","level":3,"score":0.3474000096321106},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.335999995470047},{"id":"https://openalex.org/C141353440","wikidata":"https://www.wikidata.org/wiki/Q182221","display_name":"Fuse (electrical)","level":2,"score":0.323199987411499},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.3208000063896179},{"id":"https://openalex.org/C206588197","wikidata":"https://www.wikidata.org/wiki/Q846574","display_name":"Reuse","level":2,"score":0.30709999799728394},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.2888000011444092},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.2888000011444092},{"id":"https://openalex.org/C69744172","wikidata":"https://www.wikidata.org/wiki/Q860822","display_name":"Image fusion","level":3,"score":0.2743000090122223},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.27149999141693115},{"id":"https://openalex.org/C136085584","wikidata":"https://www.wikidata.org/wiki/Q910289","display_name":"Overlay","level":2,"score":0.2612999975681305},{"id":"https://openalex.org/C52622490","wikidata":"https://www.wikidata.org/wiki/Q1026626","display_name":"Feature extraction","level":2,"score":0.25780001282691956},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2565999925136566},{"id":"https://openalex.org/C173414695","wikidata":"https://www.wikidata.org/wiki/Q5510276","display_name":"Fusion mechanism","level":4,"score":0.25429999828338623},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.25220000743865967}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.10780","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10780","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":"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.2605.10780","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10780","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":"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":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Representation":[0,76],"autoencoders":[1],"that":[2,41,61,82],"reuse":[3],"frozen":[4,101,120],"pretrained":[5,102],"vision":[6],"encoders":[7],"as":[8,51,175],"visual":[9,43,182],"tokenizers":[10,183],"have":[11],"achieved":[12],"strong":[13],"reconstruction":[14,170],"and":[15,60,91,141,169],"generation":[16,143],"quality.":[17],"However,":[18],"existing":[19],"methods":[20],"universally":[21],"extract":[22],"features":[23,65],"from":[24,137,145],"only":[25],"the":[26,31,47,111,114,119,124,129],"last":[27,48],"encoder":[28,86],"layer,":[29],"discarding":[30],"rich":[32],"hierarchical":[33],"information":[34],"distributed":[35],"across":[36],"intermediate":[37],"layers.":[38],"We":[39,72],"show":[40],"low-level":[42],"details":[44],"survive":[45],"in":[46,188],"layer":[49],"merely":[50],"attenuated":[52],"residuals":[53],"after":[54],"multiple":[55],"layers":[56,87],"of":[57,118],"semantic":[58],"abstraction,":[59],"explicitly":[62],"fusing":[63],"multi-layer":[64],"can":[66],"substantially":[67],"recover":[68],"this":[69],"lost":[70],"information.":[71],"propose":[73],"DRoRAE":[74,134],"(Depth-Routed":[75],"AutoEncoder),":[77],"a":[78,100,161,176],"lightweight":[79],"fusion":[80,112,167],"module":[81],"adaptively":[83],"aggregates":[84],"all":[85],"via":[88],"energy-constrained":[89],"routing":[90],"incremental":[92],"correction,":[93],"producing":[94],"an":[95],"enriched":[96,130],"latent":[97],"compatible":[98],"with":[99,151],"decoder.":[103],"A":[104],"three-phase":[105],"decoupled":[106],"training":[107],"strategy":[108],"first":[109],"learns":[110],"under":[113],"implicit":[115],"distributional":[116],"constraint":[117],"decoder,":[121],"then":[122],"fine-tunes":[123],"decoder":[125],"to":[126,139,147,155,185],"fully":[127],"exploit":[128],"representation.":[131],"On":[132],"ImageNet-256,":[133],"reduces":[135],"rFID":[136],"0.57":[138],"0.29":[140],"improves":[142],"FID":[144],"1.74":[146],"1.65":[148],"(with":[149],"AutoGuidance),":[150],"gains":[152],"also":[153],"transferring":[154],"text-to-image":[156],"synthesis.":[157],"Furthermore,":[158],"we":[159],"uncover":[160],"log-linear":[162],"scaling":[163],"law":[164],"($R^2{=}0.86$)":[165],"between":[166],"capacity":[168],"quality,":[171],"identifying":[172],"\\textit{representation":[173],"richness}":[174],"new,":[177],"predictably":[178],"scalable":[179],"dimension":[180],"for":[181],"analogous":[184],"vocabulary":[186],"size":[187],"NLP.":[189]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-13T00:00:00"}