{"id":"https://openalex.org/W4415708313","doi":"https://doi.org/10.1109/icme59968.2025.11210023","title":"Reinforcement Learning-based Token Pruning in Vision Transformers: A Markov Game Approach","display_name":"Reinforcement Learning-based Token Pruning in Vision Transformers: A Markov Game Approach","publication_year":2025,"publication_date":"2025-06-30","ids":{"openalex":"https://openalex.org/W4415708313","doi":"https://doi.org/10.1109/icme59968.2025.11210023"},"language":null,"primary_location":{"id":"doi:10.1109/icme59968.2025.11210023","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icme59968.2025.11210023","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE International Conference on Multimedia and Expo (ICME)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101658258","display_name":"Chenglong Lu","orcid":"https://orcid.org/0000-0002-8262-5076"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chenglong Lu","raw_affiliation_strings":["Fudan University,School of Computer Science,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fudan University,School of Computer Science,China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103674358","display_name":"Shen Liang","orcid":"https://orcid.org/0000-0002-2246-8427"},"institutions":[{"id":"https://openalex.org/I4210091437","display_name":"Sorbonne Paris Cit\u00e9","ror":"https://ror.org/001z21q04","country_code":"FR","type":"other","lineage":["https://openalex.org/I4210091437"]},{"id":"https://openalex.org/I4403386760","display_name":"Laboratoire Informatique Paris Descartes","ror":"https://ror.org/04s80ef73","country_code":"FR","type":"facility","lineage":["https://openalex.org/I204730241","https://openalex.org/I4403386760"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"Shen Liang","raw_affiliation_strings":["Université Paris Cité,Data Intelligence Institute of Paris (diiP) & LIPADE,Paris,France"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Université Paris Cité,Data Intelligence Institute of Paris (diiP) & LIPADE,Paris,France","institution_ids":["https://openalex.org/I4210091437","https://openalex.org/I4403386760"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100684660","display_name":"Xuewei Wang","orcid":"https://orcid.org/0000-0002-5780-6553"},"institutions":[{"id":"https://openalex.org/I90090648","display_name":"Shijiazhuang Tiedao University","ror":"https://ror.org/022e9e065","country_code":"CN","type":"education","lineage":["https://openalex.org/I90090648"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xuewei Wang","raw_affiliation_strings":["Shijiazhuang Tiedao University,State Key Laboratory of Mechanical Behavior and System Safety of Traffic Engineering,Shijiazhuang,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shijiazhuang Tiedao University,State Key Laboratory of Mechanical Behavior and System Safety of Traffic Engineering,Shijiazhuang,China","institution_ids":["https://openalex.org/I90090648"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100662016","display_name":"Wei Wang","orcid":"https://orcid.org/0000-0002-4567-2834"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wei Wang","raw_affiliation_strings":["Fudan University,School of Computer Science,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fudan University,School of Computer Science,China","institution_ids":["https://openalex.org/I24943067"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"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":"1","last_page":"6"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.397599995136261,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.397599995136261,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.17839999496936798,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.07440000027418137,"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/security-token","display_name":"Security token","score":0.7303000092506409},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.6699000000953674},{"id":"https://openalex.org/keywords/markov-decision-process","display_name":"Markov decision process","score":0.6223999857902527},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.6111000180244446},{"id":"https://openalex.org/keywords/partially-observable-markov-decision-process","display_name":"Partially observable Markov decision process","score":0.4884999990463257},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.47609999775886536},{"id":"https://openalex.org/keywords/markov-chain","display_name":"Markov chain","score":0.43630000948905945}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.733299970626831},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.7303000092506409},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6699000000953674},{"id":"https://openalex.org/C106189395","wikidata":"https://www.wikidata.org/wiki/Q176789","display_name":"Markov decision process","level":3,"score":0.6223999857902527},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.6111000180244446},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5185999870300293},{"id":"https://openalex.org/C17098449","wikidata":"https://www.wikidata.org/wiki/Q176814","display_name":"Partially observable Markov decision process","level":4,"score":0.4884999990463257},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.47609999775886536},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.46889999508857727},{"id":"https://openalex.org/C98763669","wikidata":"https://www.wikidata.org/wiki/Q176645","display_name":"Markov chain","level":2,"score":0.43630000948905945},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.38679999113082886},{"id":"https://openalex.org/C159886148","wikidata":"https://www.wikidata.org/wiki/Q176645","display_name":"Markov process","level":2,"score":0.35120001435279846},{"id":"https://openalex.org/C111696304","wikidata":"https://www.wikidata.org/wiki/Q2303697","display_name":"Sorting","level":2,"score":0.3149999976158142},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.30000001192092896},{"id":"https://openalex.org/C23224414","wikidata":"https://www.wikidata.org/wiki/Q176769","display_name":"Hidden Markov model","level":2,"score":0.271699994802475},{"id":"https://openalex.org/C163836022","wikidata":"https://www.wikidata.org/wiki/Q6771326","display_name":"Markov model","level":3,"score":0.2703000009059906},{"id":"https://openalex.org/C173801870","wikidata":"https://www.wikidata.org/wiki/Q201413","display_name":"Heuristic","level":2,"score":0.25119999051094055}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icme59968.2025.11210023","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icme59968.2025.11210023","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE International Conference on Multimedia and Expo (ICME)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":7,"referenced_works":["https://openalex.org/W1542941925","https://openalex.org/W2108598243","https://openalex.org/W2951360122","https://openalex.org/W3176196997","https://openalex.org/W3188427387","https://openalex.org/W4312872526","https://openalex.org/W4382466174"],"related_works":[],"abstract_inverted_index":{"Vision":[0],"Transformers":[1],"(ViTs)":[2],"have":[3],"computational":[4],"costs":[5],"scaling":[6],"quadratically":[7],"with":[8],"the":[9,34,47,120,127],"number":[10],"of":[11,37,111,141],"tokens,":[12],"calling":[13],"for":[14,46,96],"effective":[15],"token":[16,38,67],"pruning":[17,39,64,68,94],"policies.":[18],"Most":[19],"existing":[20],"policies":[21],"are":[22],"handcrafted,":[23],"lacking":[24],"adaptivity":[25],"to":[26,32,60,114,132],"varying":[27],"inputs.":[28],"Moreover,":[29],"they":[30],"fail":[31],"consider":[33],"sequential":[35,71],"nature":[36],"across":[40],"multiple":[41],"layers.":[42],"In":[43],"this":[44],"work,":[45],"first":[48],"time":[49],"(as":[50],"far":[51],"as":[52,69,77],"we":[53,55,74],"know),":[54],"exploit":[56],"Reinforcement":[57],"Learning":[58],"(RL)":[59],"data-adaptively":[61],"learn":[62],"a":[63,70,78,97,137],"policy.":[65],"Formulating":[66],"decision-making":[72],"problem,":[73],"model":[75],"it":[76],"Markov":[79],"Game":[80],"and":[81,109,117],"utilize":[82],"Multi-Agent":[83],"Proximal":[84],"Policy":[85],"Optimization":[86],"(MAPPO)":[87],"where":[88],"each":[89],"agent":[90],"makes":[91],"an":[92],"individualized":[93],"decision":[95],"single":[98],"token.":[99],"We":[100],"also":[101],"develop":[102],"reward":[103],"functions":[104],"that":[105],"enable":[106],"simultaneous":[107],"collaboration":[108],"competition":[110],"these":[112],"agents":[113],"balance":[115],"efficiency":[116],"accuracy.":[118],"On":[119],"well-known":[121],"ImageNet-1k":[122],"dataset,":[123],"our":[124],"method":[125],"improves":[126],"inference":[128],"speed":[129],"by":[130],"up":[131],"44%":[133],"while":[134],"incurring":[135],"only":[136],"negligible":[138],"accuracy":[139],"drop":[140],"0.4%.":[142],"The":[143],"source":[144],"code":[145],"is":[146],"available":[147],"at":[148],"https://github.com/daashuai/rl4evit.":[149]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-30T00:00:00"}