{"id":"https://openalex.org/W7137832803","doi":"https://doi.org/10.1609/aaai.v40i29.39605","title":"MMG-Vid: Maximizing Marginal Gains at Segment-level and Token-level for Efficient Video LLMs","display_name":"MMG-Vid: Maximizing Marginal Gains at Segment-level and Token-level for Efficient Video LLMs","publication_year":2026,"publication_date":"2026-03-14","ids":{"openalex":"https://openalex.org/W7137832803","doi":"https://doi.org/10.1609/aaai.v40i29.39605"},"language":null,"primary_location":{"id":"doi:10.1609/aaai.v40i29.39605","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i29.39605","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.v40i29.39605","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5110929742","display_name":"Junpeng Ma","orcid":"https://orcid.org/0009-0008-9878-1218"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Junpeng Ma","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129733493","display_name":"Qizhe Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qizhe Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129709628","display_name":"Ming Lu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ming Lu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129699982","display_name":"Zhibin Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhibin Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129740375","display_name":"Qiang Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qiang Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129653548","display_name":"Jun Song","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jun Song","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5129734509","display_name":"Shanghang Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shanghang Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":7,"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":0.0477208,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"40","issue":"29","first_page":"24253","last_page":"24261"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.8305000066757202,"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":0.8305000066757202,"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":0.05350000038743019,"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":0.05169999971985817,"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/security-token","display_name":"Security token","score":0.7267000079154968},{"id":"https://openalex.org/keywords/redundancy","display_name":"Redundancy (engineering)","score":0.529699981212616},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.4945000112056732},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.4869999885559082},{"id":"https://openalex.org/keywords/frame","display_name":"Frame (networking)","score":0.4214000105857849},{"id":"https://openalex.org/keywords/budget-constraint","display_name":"Budget constraint","score":0.41179999709129333},{"id":"https://openalex.org/keywords/marginal-cost","display_name":"Marginal cost","score":0.37560001015663147}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7415000200271606},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.7267000079154968},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.529699981212616},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.4945000112056732},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.4869999885559082},{"id":"https://openalex.org/C126042441","wikidata":"https://www.wikidata.org/wiki/Q1324888","display_name":"Frame (networking)","level":2,"score":0.4214000105857849},{"id":"https://openalex.org/C8505890","wikidata":"https://www.wikidata.org/wiki/Q605095","display_name":"Budget constraint","level":2,"score":0.41179999709129333},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4081000089645386},{"id":"https://openalex.org/C66887028","wikidata":"https://www.wikidata.org/wiki/Q382444","display_name":"Marginal cost","level":2,"score":0.37560001015663147},{"id":"https://openalex.org/C2780186347","wikidata":"https://www.wikidata.org/wiki/Q11414","display_name":"Subnetwork","level":2,"score":0.3662000000476837},{"id":"https://openalex.org/C138551086","wikidata":"https://www.wikidata.org/wiki/Q842271","display_name":"Marginal utility","level":2,"score":0.325300008058548},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.323199987411499},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3199999928474426},{"id":"https://openalex.org/C2777021972","wikidata":"https://www.wikidata.org/wiki/Q22976830","display_name":"Uniqueness","level":2,"score":0.3125},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.31209999322891235},{"id":"https://openalex.org/C3261483","wikidata":"https://www.wikidata.org/wiki/Q119565","display_name":"Frame rate","level":2,"score":0.28060001134872437},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.2759999930858612},{"id":"https://openalex.org/C202474056","wikidata":"https://www.wikidata.org/wiki/Q1931635","display_name":"Video tracking","level":3,"score":0.25270000100135803}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1609/aaai.v40i29.39605","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i29.39605","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"}],"best_oa_location":{"id":"doi:10.1609/aaai.v40i29.39605","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i29.39605","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":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Video":[0],"Large":[1],"Language":[2],"Models":[3],"(VLLMs)":[4],"excel":[5],"in":[6],"video":[7,45,50,87],"understanding,":[8],"but":[9],"their":[10],"excessive":[11],"visual":[12,30,66,173],"tokens":[13,174],"pose":[14],"a":[15,53,63,115],"significant":[16],"computational":[17],"challenge":[18],"for":[19,101],"real-world":[20],"applications.":[21],"Current":[22],"methods":[23],"aim":[24],"to":[25,104],"enhance":[26],"inference":[27],"efficiency":[28,151],"by":[29,73,180],"token":[31,67,99,147],"pruning.":[32],"However,":[33],"they":[34,48],"do":[35],"not":[36],"consider":[37],"the":[38,86,98,106,129,142,145,166,177],"dynamic":[39],"characteristics":[40],"and":[41,80,94,124,175],"temporal":[42],"dependencies":[43],"of":[44,109,132,144,165],"frames,":[46],"as":[47,52],"perceive":[49],"understanding":[51],"multi-frame":[54],"task.":[55],"To":[56],"address":[57],"these":[58],"challenges,":[59],"we":[60,83,113],"propose":[61,114],"MMG-Vid,":[62],"novel":[64],"training-free":[65],"pruning":[68],"framework":[69],"that":[70,119,159],"removes":[71],"redundancy":[72],"Maximizing":[74],"Marginal":[75],"Gains":[76],"at":[77],"both":[78,137],"segment-level":[79],"token-level.":[81],"Specifically,":[82],"first":[84],"divide":[85],"into":[88],"segments":[89],"based":[90],"on":[91,182],"frame":[92],"similarity,":[93],"then":[95],"dynamically":[96],"allocate":[97],"budget":[100],"each":[102,110,133],"segment":[103],"maximize":[105,141],"marginal":[107,130],"gain":[108,131],"segment.":[111],"Subsequently,":[112],"temporal-guided":[116],"DPC":[117],"algorithm":[118],"jointly":[120],"models":[121],"inter-frame":[122],"uniqueness":[123],"intra-frame":[125],"diversity,":[126],"thereby":[127],"maximizing":[128],"token.":[134],"By":[135],"combining":[136],"stages,":[138],"MMG-Vid":[139,160],"can":[140,161],"utilization":[143],"limited":[146],"budget,":[148],"significantly":[149],"improving":[150],"while":[152,169],"maintaining":[153],"strong":[154],"performance.":[155],"Extensive":[156],"experiments":[157],"demonstrate":[158],"maintain":[162],"over":[163],"99.5%":[164],"original":[167],"performance,":[168],"effectively":[170],"reducing":[171],"75%":[172],"accelerating":[176],"prefilling":[178],"stage":[179],"3.9x":[181],"LLaVA-OneVision-7B.":[183]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-03-18T00:00:00"}