{"id":"https://openalex.org/W7204839614","doi":"https://doi.org/10.1145/3767308.3835088","title":"Token-Budget Distillation: Transferring Full-Token Semantics to Compressed Video Vision-Language Models","display_name":"Token-Budget Distillation: Transferring Full-Token Semantics to Compressed Video Vision-Language Models","publication_year":2026,"publication_date":"2026-11-06","ids":{"openalex":"https://openalex.org/W7204839614","doi":"https://doi.org/10.1145/3767308.3835088"},"language":"en","primary_location":{"id":"doi:10.1145/3767308.3835088","doi":"https://doi.org/10.1145/3767308.3835088","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3767308.3835088","pdf_url":null,"source":{"id":"https://openalex.org/S4363608757","display_name":"ACM International Conference on Multimedia (ACM MM)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 34th ACM International Conference on Multimedia","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://doi.org/10.1145/3767308.3835088","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5121178688","display_name":"Xiaoyang Guo","orcid":"https://orcid.org/0009-0001-0848-1369","observed_orcids":["https://orcid.org/0009-0001-0848-1369"]},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xiaoyang Guo","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, Guangdong, China"],"raw_orcid":"https://orcid.org/0009-0001-0848-1369","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, Guangdong, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143406522","display_name":"Guoping Luo","orcid":"https://orcid.org/0009-0003-8283-6273","observed_orcids":["https://orcid.org/0009-0003-8283-6273"]},"institutions":[{"id":"https://openalex.org/I141945490","display_name":"University of British Columbia","ror":"https://ror.org/03rmrcq20","country_code":"CA","type":"education","lineage":["https://openalex.org/I141945490"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Guoping Luo","raw_affiliation_strings":["The University of British Columbia, Vancouver, British Columbia, Canada"],"raw_orcid":"https://orcid.org/0009-0003-8283-6273","affiliations":[{"raw_affiliation_string":"The University of British Columbia, Vancouver, British Columbia, Canada","institution_ids":["https://openalex.org/I141945490"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125528192","display_name":"Jusheng Zhang","orcid":"https://orcid.org/0009-0004-5743-2936","observed_orcids":["https://orcid.org/0009-0004-5743-2936"]},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jusheng Zhang","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, Guangdong, China"],"raw_orcid":"https://orcid.org/0009-0004-5743-2936","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, Guangdong, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123834741","display_name":"Keze Wang","orcid":"https://orcid.org/0000-0002-7817-8306","observed_orcids":["https://orcid.org/0000-0002-7817-8306"]},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Keze Wang","raw_affiliation_strings":["Sun Yat-sen University, Guangzhou, Guangdong, China"],"raw_orcid":"https://orcid.org/0000-0002-7817-8306","affiliations":[{"raw_affiliation_string":"Sun Yat-sen University, Guangzhou, Guangdong, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5137910713","display_name":"Wenhao Wang","orcid":"https://orcid.org/0000-0001-8727-1572","observed_orcids":["https://orcid.org/0000-0001-8727-1572"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wenhao Wang","raw_affiliation_strings":["Vast Intelligence Lab, Sydney, New South Wales, Australia"],"raw_orcid":"https://orcid.org/0000-0001-8727-1572","affiliations":[{"raw_affiliation_string":"Vast Intelligence Lab, Sydney, New South Wales, Australia","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.9228,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.85394795,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":91,"max":95},"biblio":{"volume":null,"issue":null,"first_page":"2634","last_page":"2643"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9990000128746033,"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":0.9990000128746033,"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":0.0003000000142492354,"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":0.0003000000142492354,"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/visual-token-compression","display_name":"visual token compression","score":0.9259999990463257},{"id":"https://openalex.org/keywords/knowledge-distillation","display_name":"knowledge distillation","score":0.9459999799728394},{"id":"https://openalex.org/keywords/parameter-efficient-fine-tuning","display_name":"parameter-efficient fine-tuning","score":0.921999990940094},{"id":"https://openalex.org/keywords/lora-adapters","display_name":"LoRA adapters","score":0.9139999747276306},{"id":"https://openalex.org/keywords/semantic-drift","display_name":"semantic drift","score":0.6299999952316284},{"id":"https://openalex.org/keywords/video-understanding","display_name":"video understanding","score":0.7450000047683716},{"id":"https://openalex.org/keywords/qwen3-vl","display_name":"Qwen3-VL","score":0.9700000286102295}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7997999787330627},{"id":"https://openalex.org/C774472","wikidata":"https://www.wikidata.org/wiki/Q6760393","display_name":"Margin (machine learning)","level":2,"score":0.6664999723434448},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6589000225067139},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.6557000279426575},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.5723999738693237},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.558899998664856},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.47760000824928284},{"id":"https://openalex.org/C202474056","wikidata":"https://www.wikidata.org/wiki/Q1931635","display_name":"Video tracking","level":3,"score":0.4309999942779541},{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.4198000133037567},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.4092000126838684},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.37549999356269836},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.36410000920295715},{"id":"https://openalex.org/C106030495","wikidata":"https://www.wikidata.org/wiki/Q1797012","display_name":"Video compression picture types","level":4,"score":0.3562999963760376},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.35339999198913574},{"id":"https://openalex.org/C202708506","wikidata":"https://www.wikidata.org/wiki/Q7449050","display_name":"Semantic compression","level":5,"score":0.31679999828338623},{"id":"https://openalex.org/C25797200","wikidata":"https://www.wikidata.org/wiki/Q828137","display_name":"Compression ratio","level":3,"score":0.3050000071525574},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.30469998717308044},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.2985000014305115},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.28700000047683716},{"id":"https://openalex.org/C65483669","wikidata":"https://www.wikidata.org/wiki/Q3536669","display_name":"Video processing","level":2,"score":0.28029999136924744},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.2791000008583069},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.26750001311302185},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.2563000023365021},{"id":"https://openalex.org/C67277372","wikidata":"https://www.wikidata.org/wiki/Q7449085","display_name":"Semantic role labeling","level":3,"score":0.25189998745918274}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1145/3767308.3835088","doi":"https://doi.org/10.1145/3767308.3835088","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3767308.3835088","pdf_url":null,"source":{"id":"https://openalex.org/S4363608757","display_name":"ACM International Conference on Multimedia (ACM MM)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 34th ACM International Conference on Multimedia","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2608.28138","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.28138","pdf_url":"https://arxiv.org/pdf/2608.28138","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.28138","doi":"https://doi.org/10.48550/arxiv.2608.28138","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.28138","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,"listed_in":[],"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"}],"best_oa_location":{"id":"doi:10.1145/3767308.3835088","doi":"https://doi.org/10.1145/3767308.3835088","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3767308.3835088","pdf_url":null,"source":{"id":"https://openalex.org/S4363608757","display_name":"ACM International Conference on Multimedia (ACM MM)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 34th ACM International Conference on Multimedia","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G1099008978","display_name":null,"funder_award_id":"23hytd006","funder_id":"https://openalex.org/F4320321160","funder_display_name":"Sun Yat-sen University"},{"id":"https://openalex.org/G327634725","display_name":null,"funder_award_id":"62276283","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320321160","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71"},{"id":"https://openalex.org/F4320321569","display_name":"China Meteorological Administration","ror":"https://ror.org/00bx3rb98"},{"id":"https://openalex.org/F4320335787","display_name":"Fundamental Research Funds for the Central Universities","ror":null}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7204839614.pdf","grobid_xml":"https://content.openalex.org/works/W7204839614.grobid-xml"},"referenced_works_count":43,"referenced_works":["https://openalex.org/W1821462560","https://openalex.org/W2978017171","https://openalex.org/W3152956381","https://openalex.org/W3174770825","https://openalex.org/W4306886919","https://openalex.org/W4330338093","https://openalex.org/W4393212018","https://openalex.org/W4396818380","https://openalex.org/W4399655895","https://openalex.org/W4400488577","https://openalex.org/W4402671548","https://openalex.org/W4402702917","https://openalex.org/W4402727142","https://openalex.org/W4402754238","https://openalex.org/W4403346092","https://openalex.org/W4403370767","https://openalex.org/W4403853618","https://openalex.org/W4404261828","https://openalex.org/W4404784276","https://openalex.org/W4407386187","https://openalex.org/W4412887841","https://openalex.org/W4413145840","https://openalex.org/W4413146669","https://openalex.org/W4413146710","https://openalex.org/W4413156814","https://openalex.org/W4413158337","https://openalex.org/W4416035093","https://openalex.org/W4416045996","https://openalex.org/W4416288898","https://openalex.org/W4416600567","https://openalex.org/W4416611614","https://openalex.org/W4416780398","https://openalex.org/W4417283153","https://openalex.org/W7108068597","https://openalex.org/W7111360605","https://openalex.org/W7114769457","https://openalex.org/W7114808432","https://openalex.org/W7125827338","https://openalex.org/W7128501944","https://openalex.org/W7128773051","https://openalex.org/W7160165061","https://openalex.org/W7160187380","https://openalex.org/W7166493911"],"related_works":["https://openalex.org/W7172394348","https://openalex.org/W7161825108","https://openalex.org/W7151453520","https://openalex.org/W7215811224","https://openalex.org/W7131409469","https://openalex.org/W7140223168","https://openalex.org/W4416037134","https://openalex.org/W4416035970","https://openalex.org/W7214988040","https://openalex.org/W4416399887"],"abstract_inverted_index":{"Adapting":[0],"video":[1,9,56,80,146,156],"vision-language":[2],"models":[3],"(VLMs)":[4],"is":[5,105],"computationally":[6],"expensive":[7],"because":[8],"inputs":[10,36],"produce":[11],"a":[12,50,59,90,95,102],"large":[13],"number":[14],"of":[15,130,179,194],"visual":[16,25,75],"tokens,":[17],"making":[18],"both":[19,165],"fine-tuning":[20,52],"and":[21,41,72,101,116,152,167,196],"inference":[22],"costly.":[23],"Although":[24],"token":[26,61,76,139],"compression":[27,77],"can":[28],"reduce":[29],"this":[30],"overhead,":[31],"direct":[32],"adaptation":[33],"on":[34,144,185],"compressed":[35,103],"often":[37],"causes":[38],"semantic":[39,128],"drift":[40],"noticeable":[42],"performance":[43],"degradation.":[44],"We":[45,141],"present":[46],"Token-Budget":[47],"Distillation":[48],"(TBD),":[49],"parameter-efficient":[51],"framework":[53],"for":[54],"adapting":[55],"VLMs":[57],"under":[58,86,137,164],"fixed":[60],"budget.":[62],"TBD":[63,88,143,159,176],"freezes":[64],"the":[65,79,123,127,131,180],"pretrained":[66],"backbone,":[67],"updates":[68],"only":[69],"LoRA":[70],"adapters,":[71],"integrates":[73],"FlashVID-based":[74],"into":[78],"pathway.":[81],"To":[82],"preserve":[83],"full-token":[84,96,132],"semantics":[85],"compression,":[87],"employs":[89],"dual-path":[91],"teacher-student":[92],"design,":[93],"where":[94],"teacher":[97],"provides":[98],"stable":[99],"supervision":[100],"student":[104,124],"optimized":[106],"with":[107],"task":[108],"loss,":[109],"answer-region":[110],"KL":[111],"distillation,":[112,115],"GT-anchored":[113],"margin":[114],"reliability-aware":[117],"KD":[118],"control.":[119],"This":[120],"design":[121],"enables":[122],"to":[125],"recover":[126],"behavior":[129],"model":[133],"while":[134],"remaining":[135],"efficient":[136],"aggressive":[138,168],"reduction.":[140],"evaluate":[142],"three":[145],"VLM":[147],"backbones,":[148],"including":[149],"LLaVA-Video,":[150],"LLaVA-OneVision,":[151],"Qwen3-VL-8B-Instruct,":[153],"across":[154],"four":[155],"understanding":[157],"benchmarks.":[158],"consistently":[160],"outperforms":[161],"compression-only":[162],"baselines":[163],"moderate":[166],"compression.":[169],"On":[170],"LLaVA-Video":[171],"at":[172,187],"retention":[173],"ratio":[174],"R=10%,":[175,188],"preserves":[177],"97.0%":[178],"Vanilla":[181],"model's":[182],"average":[183,192],"accuracy;":[184],"LLaVA-OneVision":[186],"it":[189],"achieves":[190],"an":[191],"score":[193],"58.4":[195],"matches":[197],"100.0%":[198],"relative":[199],"accuracy.":[200]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-10T11:39:43.021295","created_date":"2026-09-01T00:00:00"}