{"id":"https://openalex.org/W7133361807","doi":"https://doi.org/10.48550/arxiv.2603.00188","title":"Efficient Long-Horizon GUI Agents via Training-Free KV Cache Compression","display_name":"Efficient Long-Horizon GUI Agents via Training-Free KV Cache Compression","publication_year":2026,"publication_date":"2026-02-27","ids":{"openalex":"https://openalex.org/W7133361807","doi":"https://doi.org/10.48550/arxiv.2603.00188"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.00188","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.00188","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":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.2603.00188","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5127917806","display_name":"Bowen Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Bowen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128030405","display_name":"Zhou Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127891961","display_name":"Wanli Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Wanli","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128016331","display_name":"Jingyu Xiao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiao, Jingyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5127896827","display_name":"Haoqian Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Haoqian","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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.7150999903678894,"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":0.7150999903678894,"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":0.10140000283718109,"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/T11605","display_name":"Visual Attention and Saliency Detection","score":0.0340999998152256,"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/cache","display_name":"Cache","score":0.6044999957084656},{"id":"https://openalex.org/keywords/redundancy","display_name":"Redundancy (engineering)","score":0.49900001287460327},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.45649999380111694},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.38359999656677246},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.3758000135421753},{"id":"https://openalex.org/keywords/software-deployment","display_name":"Software deployment","score":0.3467000126838684},{"id":"https://openalex.org/keywords/memory-footprint","display_name":"Memory footprint","score":0.33169999718666077},{"id":"https://openalex.org/keywords/cache-algorithms","display_name":"Cache algorithms","score":0.3018999993801117}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.873199999332428},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.6044999957084656},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.49900001287460327},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.45649999380111694},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.38359999656677246},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.3758000135421753},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.3686999976634979},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.3467000126838684},{"id":"https://openalex.org/C74912251","wikidata":"https://www.wikidata.org/wiki/Q6815727","display_name":"Memory footprint","level":2,"score":0.33169999718666077},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.32010000944137573},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.31839999556541443},{"id":"https://openalex.org/C38556500","wikidata":"https://www.wikidata.org/wiki/Q13404475","display_name":"Cache algorithms","level":4,"score":0.3018999993801117},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.2939999997615814},{"id":"https://openalex.org/C189783530","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"CPU cache","level":3,"score":0.2906999886035919},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.28349998593330383},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.28209999203681946},{"id":"https://openalex.org/C25536678","wikidata":"https://www.wikidata.org/wiki/Q5015977","display_name":"Cache invalidation","level":5,"score":0.2782000005245209},{"id":"https://openalex.org/C101468663","wikidata":"https://www.wikidata.org/wiki/Q1620158","display_name":"Modular design","level":2,"score":0.2775000035762787},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.26589998602867126},{"id":"https://openalex.org/C7545210","wikidata":"https://www.wikidata.org/wiki/Q838123","display_name":"Data redundancy","level":2,"score":0.2628999948501587},{"id":"https://openalex.org/C189930140","wikidata":"https://www.wikidata.org/wiki/Q1112878","display_name":"CAS latency","level":4,"score":0.26170000433921814},{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.25760000944137573},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.25619998574256897},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.251800000667572}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.00188","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.00188","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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2603.00188","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.00188","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":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":{"Large":[0],"Vision-Language":[1],"Models":[2],"(VLMs)":[3],"have":[4,39],"emerged":[5],"as":[6],"powerful":[7],"engines":[8],"for":[9,42,95,189],"autonomous":[10],"GUI":[11,54,71,97,107,191],"agents,":[12],"yet":[13],"their":[14],"deployment":[15],"is":[16],"severely":[17],"constrained":[18],"by":[19,82,137,147],"the":[20,27,102,130,155],"substantial":[21],"memory":[22],"footprint":[23],"and":[24,122],"latency":[25],"of":[26,133],"Key-Value":[28],"(KV)":[29],"cache":[30,36,91,166],"during":[31],"long-horizon":[32],"interactions.":[33],"While":[34],"existing":[35],"compression":[37,92],"methods":[38],"proven":[40],"effective":[41],"LLMs,":[43],"we":[44,85],"empirically":[45],"demonstrate":[46,160],"that":[47,99,161],"they":[48],"suffer":[49],"from":[50],"suboptimal":[51],"performance":[52,180],"in":[53],"scenarios":[55],"due":[56],"to":[57,182],"a":[58,88,112,164,170,186],"fundamental":[59],"misalignment:":[60],"unlike":[61],"general":[62],"visual":[63],"tasks":[64],"where":[65],"attention":[66,72],"sparsity":[67],"varies":[68],"across":[69,77],"layers,":[70],"patterns":[73],"exhibit":[74],"uniform":[75],"high-sparsity":[76],"all":[78],"transformer":[79],"layers.":[80],"Motivated":[81],"this":[83],"insight,":[84],"propose":[86],"ST-Lite,":[87],"training-free":[89],"KV":[90,152],"framework":[93],"tailored":[94],"efficient":[96],"agents":[98],"explicitly":[100],"addresses":[101],"dynamic":[103],"spatio-trajectory":[104],"dependencies":[105],"within":[106,154],"data":[108],"streams.":[109],"ST-Lite":[110,168],"introduces":[111],"novel":[113],"dual-branch":[114],"scoring":[115],"policy":[116],"incorporating":[117],"Component-centric":[118],"Spatial":[119],"Saliency":[120],"(CSS)":[121],"Trajectory-aware":[123],"Semantic":[124],"Gating":[125],"(TSG).":[126],"Specifically,":[127],"CSS":[128],"preserves":[129],"structural":[131],"integrity":[132],"interactive":[134],"UI":[135],"elements":[136],"evaluating":[138],"local":[139],"neighborhood":[140],"saliency,":[141],"while":[142,174],"TSG":[143],"mitigates":[144],"historical":[145],"redundancy":[146],"dynamically":[148],"filtering":[149],"visually":[150],"repetitive":[151],"pairs":[153],"interaction":[156],"trajectory.":[157],"Extensive":[158],"evaluations":[159],"with":[162],"only":[163],"10-20%":[165],"budget,":[167],"achieves":[169],"2.45x":[171],"decoding":[172],"acceleration":[173],"maintaining":[175],"comparable":[176],"or":[177],"even":[178],"superior":[179],"compared":[181],"full-cache":[183],"baselines,":[184],"offering":[185],"scalable":[187],"solution":[188],"resource-constrained":[190],"agents.":[192]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-03-04T00:00:00"}