{"id":"https://openalex.org/W7221069330","doi":"https://doi.org/10.3390/ai7100413","title":"When Deployment Changes a Learned Action: Post-Inference Projection in Offline Multi-Agent Reinforcement Learning","display_name":"When Deployment Changes a Learned Action: Post-Inference Projection in Offline Multi-Agent Reinforcement Learning","publication_year":2026,"publication_date":"2026-10-08","ids":{"openalex":"https://openalex.org/W7221069330","doi":"https://doi.org/10.3390/ai7100413"},"language":"en","primary_location":{"id":"doi:10.3390/ai7100413","doi":"https://doi.org/10.3390/ai7100413","is_oa":true,"landing_page_url":"https://doi.org/10.3390/ai7100413","pdf_url":null,"source":{"id":"https://openalex.org/S4210200511","display_name":"AI","issn_l":"2673-2688","issn":["2673-2688"],"is_oa":true,"is_in_doaj":true,"is_core":true,"listed_in":["cwts-core","doaj","jufo-1","norway-1"],"host_organization":"https://openalex.org/P4310310987","host_organization_name":"Multidisciplinary Digital Publishing Institute","host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"AI","raw_type":"journal-article"},"volume":null,"type":"article","indexed_in":["crossref","doaj"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.3390/ai7100413","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5014044631","display_name":"Abzal E. Kyzyrkanov","orcid":"https://orcid.org/0000-0002-8817-8617","observed_orcids":["https://orcid.org/0000-0002-8817-8617"]},"institutions":[{"id":"https://openalex.org/I4405257690","display_name":"Astana IT University","ror":"https://ror.org/05jnvbc31","country_code":"KZ","type":"education","lineage":["https://openalex.org/I4405257690"]}],"countries":["KZ"],"is_corresponding":true,"raw_author_name":"Abzal E. Kyzyrkanov","raw_affiliation_strings":["School of Software Engineering, Astana IT University, Astana 020000, Kazakhstan"],"raw_orcid":"https://orcid.org/0000-0002-8817-8617","affiliations":[{"raw_affiliation_string":"School of Software Engineering, Astana IT University, Astana 020000, Kazakhstan","institution_ids":["https://openalex.org/I4405257690"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5003517480","display_name":"Zhenis Otarbay","orcid":"https://orcid.org/0000-0003-4111-0719","observed_orcids":["https://orcid.org/0000-0003-4111-0719"]},"institutions":[{"id":"https://openalex.org/I4405257690","display_name":"Astana IT University","ror":"https://ror.org/05jnvbc31","country_code":"KZ","type":"education","lineage":["https://openalex.org/I4405257690"]},{"id":"https://openalex.org/I60559429","display_name":"Nazarbayev University","ror":"https://ror.org/052bx8q98","country_code":"KZ","type":"education","lineage":["https://openalex.org/I60559429"]}],"countries":["KZ"],"is_corresponding":true,"raw_author_name":"Zhenis Otarbay","raw_affiliation_strings":["Department of Computer Science, School of Engineering and Digital Sciences, Nazarbayev University, Astana 010000, Kazakhstan","School of Software Engineering, Astana IT University, Astana 020000, Kazakhstan"],"raw_orcid":"https://orcid.org/0000-0003-4111-0719","affiliations":[{"raw_affiliation_string":"Department of Computer Science, School of Engineering and Digital Sciences, Nazarbayev University, Astana 010000, Kazakhstan","institution_ids":["https://openalex.org/I60559429"]},{"raw_affiliation_string":"School of Software Engineering, Astana IT University, Astana 020000, Kazakhstan","institution_ids":["https://openalex.org/I4405257690"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066383967","display_name":"Yedil S. Nurakhov","orcid":"https://orcid.org/0000-0003-0799-7555","observed_orcids":["https://orcid.org/0000-0003-0799-7555"]},"institutions":[{"id":"https://openalex.org/I185571130","display_name":"Al-Farabi Kazakh National University","ror":"https://ror.org/03q0vrn42","country_code":"KZ","type":"education","lineage":["https://openalex.org/I185571130"]}],"countries":["KZ"],"is_corresponding":false,"raw_author_name":"Yedil S. Nurakhov","raw_affiliation_strings":["Department of Computer Sciences, Faculty of Information Technologies, Al-Farabi Kazakh National University, Almaty 050040, Kazakhstan"],"raw_orcid":"https://orcid.org/0000-0003-0799-7555","affiliations":[{"raw_affiliation_string":"Department of Computer Sciences, Faculty of Information Technologies, Al-Farabi Kazakh National University, Almaty 050040, Kazakhstan","institution_ids":["https://openalex.org/I185571130"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5049245535","display_name":"Danil V. Lebedev","orcid":"https://orcid.org/0000-0002-5186-6483","observed_orcids":["https://orcid.org/0000-0002-5186-6483"]},"institutions":[{"id":"https://openalex.org/I185571130","display_name":"Al-Farabi Kazakh National University","ror":"https://ror.org/03q0vrn42","country_code":"KZ","type":"education","lineage":["https://openalex.org/I185571130"]}],"countries":["KZ"],"is_corresponding":false,"raw_author_name":"Danil V. Lebedev","raw_affiliation_strings":["Department of Computer Sciences, Faculty of Information Technologies, Al-Farabi Kazakh National University, Almaty 050040, Kazakhstan"],"raw_orcid":"https://orcid.org/0000-0002-5186-6483","affiliations":[{"raw_affiliation_string":"Department of Computer Sciences, Faculty of Information Technologies, Al-Farabi Kazakh National University, Almaty 050040, Kazakhstan","institution_ids":["https://openalex.org/I185571130"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5003517480","https://openalex.org/A5014044631"],"corresponding_institution_ids":["https://openalex.org/I4405257690","https://openalex.org/I60559429"],"apc_list":{"value":1600,"currency":"CHF","value_usd":1930},"apc_paid":{"value":1600,"currency":"CHF","value_usd":1930},"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.78630617,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"7","issue":"10","first_page":"413","last_page":"413"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10462","display_name":"Reinforcement Learning in Robotics","score":0.9672999978065491,"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"}},"topics":[{"id":"https://openalex.org/T10462","display_name":"Reinforcement Learning in Robotics","score":0.9672999978065491,"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/T10138","display_name":"Network Traffic and Congestion Control","score":0.01080000028014183,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T10714","display_name":"Software-Defined Networks and 5G","score":0.008500000461935997,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/policy-deployment","display_name":"policy deployment","score":0.6850000023841858},{"id":"https://openalex.org/keywords/continuous-action-spaces","display_name":"continuous action spaces","score":0.49000000953674316},{"id":"https://openalex.org/keywords/network-traffic","display_name":"network traffic","score":0.34200000762939453},{"id":"https://openalex.org/keywords/policy-evaluation","display_name":"policy evaluation","score":0.574999988079071}],"concepts":[{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.7283999919891357},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.6671000123023987},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6215000152587891},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5625},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.4738999903202057},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.43459999561309814},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.3962000012397766},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.3855000138282776},{"id":"https://openalex.org/C67203356","wikidata":"https://www.wikidata.org/wiki/Q1321905","display_name":"Reinforcement","level":2,"score":0.3662000000476837},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.35740000009536743},{"id":"https://openalex.org/C199845137","wikidata":"https://www.wikidata.org/wiki/Q145490","display_name":"Network topology","level":2,"score":0.3312000036239624},{"id":"https://openalex.org/C57493831","wikidata":"https://www.wikidata.org/wiki/Q3134666","display_name":"Projection (relational algebra)","level":2,"score":0.3156000077724457},{"id":"https://openalex.org/C62354387","wikidata":"https://www.wikidata.org/wiki/Q875399","display_name":"Boundary (topology)","level":2,"score":0.3140000104904175},{"id":"https://openalex.org/C62438384","wikidata":"https://www.wikidata.org/wiki/Q331350","display_name":"Simplex","level":2,"score":0.3131999969482422},{"id":"https://openalex.org/C8642999","wikidata":"https://www.wikidata.org/wiki/Q4171168","display_name":"Hyperparameter","level":2,"score":0.3050999939441681},{"id":"https://openalex.org/C2780598303","wikidata":"https://www.wikidata.org/wiki/Q65921492","display_name":"Flexibility (engineering)","level":2,"score":0.30079999566078186},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.29829999804496765},{"id":"https://openalex.org/C44154836","wikidata":"https://www.wikidata.org/wiki/Q45045","display_name":"Simulation","level":1,"score":0.2921999990940094},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.2816999852657318},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2800999879837036},{"id":"https://openalex.org/C12590798","wikidata":"https://www.wikidata.org/wiki/Q3933199","display_name":"Replication (statistics)","level":2,"score":0.2599000036716461},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.2596000134944916},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.25209999084472656}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.3390/ai7100413","doi":"https://doi.org/10.3390/ai7100413","is_oa":true,"landing_page_url":"https://doi.org/10.3390/ai7100413","pdf_url":null,"source":{"id":"https://openalex.org/S4210200511","display_name":"AI","issn_l":"2673-2688","issn":["2673-2688"],"is_oa":true,"is_in_doaj":true,"is_core":true,"listed_in":["cwts-core","doaj","jufo-1","norway-1"],"host_organization":"https://openalex.org/P4310310987","host_organization_name":"Multidisciplinary Digital Publishing Institute","host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"AI","raw_type":"journal-article"},{"id":"pmh:oai:doaj.org/article:f5552dad3d5f496e8bc4cb5fabead339","doi":null,"is_oa":false,"landing_page_url":"https://doaj.org/article/f5552dad3d5f496e8bc4cb5fabead339","pdf_url":null,"source":{"id":"https://openalex.org/S4306401280","display_name":"DOAJ (DOAJ: Directory of Open Access Journals)","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":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"AI, Vol 7, Iss 10, p 413 (2026)","raw_type":"article"}],"best_oa_location":{"id":"doi:10.3390/ai7100413","doi":"https://doi.org/10.3390/ai7100413","is_oa":true,"landing_page_url":"https://doi.org/10.3390/ai7100413","pdf_url":null,"source":{"id":"https://openalex.org/S4210200511","display_name":"AI","issn_l":"2673-2688","issn":["2673-2688"],"is_oa":true,"is_in_doaj":true,"is_core":true,"listed_in":["cwts-core","doaj","jufo-1","norway-1"],"host_organization":"https://openalex.org/P4310310987","host_organization_name":"Multidisciplinary Digital Publishing Institute","host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"AI","raw_type":"journal-article"},"sustainable_development_goals":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":17,"referenced_works":["https://openalex.org/W2087178199","https://openalex.org/W2136451165","https://openalex.org/W2147118406","https://openalex.org/W2165021839","https://openalex.org/W2890706721","https://openalex.org/W2902423855","https://openalex.org/W3019817166","https://openalex.org/W3078093847","https://openalex.org/W3108110215","https://openalex.org/W3109937587","https://openalex.org/W3147460469","https://openalex.org/W4402217630","https://openalex.org/W4416458493","https://openalex.org/W7160027847","https://openalex.org/W7160929607","https://openalex.org/W7165924974","https://openalex.org/W7171875541"],"related_works":["https://openalex.org/W7157132189","https://openalex.org/W7214662583","https://openalex.org/W3128716542","https://openalex.org/W3213851351","https://openalex.org/W3139766052","https://openalex.org/W4324297891","https://openalex.org/W7220811260","https://openalex.org/W7160117849","https://openalex.org/W4404391400","https://openalex.org/W2296631966"],"abstract_inverted_index":{"Learned":[0],"control":[1,68],"is":[2,47,62,213],"commonly":[3],"evaluated":[4,63],"at":[5,43],"the":[6,53,96,100,105,135,187,194,207,211,242,245,255,269,277,281],"policy":[7,74,278],"output,":[8],"although":[9],"a":[10,153,198],"deployment":[11,84,246,257],"layer":[12],"may":[13],"transform":[14],"that":[15,31],"output":[16,279],"before":[17],"execution.":[18],"We":[19],"study":[20],"this":[21],"post-inference":[22],"semantic":[23],"boundary":[24],"for":[25,150],"offline":[26],"multi-agent":[27],"reinforcement":[28],"learning":[29],"policies":[30],"produce":[32,259],"three-component":[33],"simplex":[34],"actions.":[35],"Training":[36],"and":[37,76,90,104,113,125,141,178,232,254,280],"model":[38],"weights":[39],"are":[40,134],"held":[41],"fixed;":[42],"deployment,":[44],"an":[45,160],"action":[46,98,235,252],"either":[48],"preserved":[49],"or":[50],"projected":[51,208],"to":[52],"one-hot":[54],"vector":[55],"of":[56,157,193,244],"its":[57],"largest":[58],"component.":[59],"The":[60,81,131],"intervention":[61],"in":[64,109,118,171,175,180],"programmable":[65],"multipath":[66],"network":[67,71],"using":[69],"three":[70],"structures,":[72],"two":[73,256],"architectures,":[75],"ten":[77,132,147],"matched":[78],"model\u2013traffic":[79],"replications.":[80],"resulting":[82],"120":[83],"episodes":[85],"form":[86],"60":[87,111],"policy-level":[88],"contrasts":[89,112],"30":[91],"shared":[92],"topology\u2013traffic":[93],"blocks.":[94],"Preserving":[95],"continuous":[97],"favored":[99,148],"end-host":[101],"successful-session":[102,176],"fraction":[103],"controller-side":[106],"path-sum":[107],"utilization":[108],"all":[110,146],"externally":[114],"measured":[115],"traffic":[116],"loss":[117],"59;":[119],"reward,":[120],"bottleneck":[121],"utilization,":[122],"receiver":[123],"throughput,":[124],"controller-probe":[126],"latency":[127],"were":[128,167],"less":[129],"uniform.":[130],"pairs":[133],"independent":[136],"replication":[137],"unit,":[138],"with":[139,223,241],"topology":[140],"architecture":[142],"as":[143,268],"crossed":[144],"conditions;":[145],"preservation":[149],"every":[151],"outcome,":[152],"two-sided":[154],"sign-test":[155],"probability":[156],"0.002":[158],"under":[159,227],"equal-sign":[161],"reference":[162],"model.":[163],"Mean":[164],"favorable":[165],"differences":[166],"0.540":[168],"percentage":[169],"points":[170],"external":[172],"loss,":[173],"0.00539":[174],"fraction,":[177],"0.0416":[179],"reward.":[181],"Two":[182],"supplementary":[183],"online":[184],"arms":[185],"sharpen":[186],"interpretation:":[188],"top-2":[189],"renormalization":[190],"recovered":[191],"most":[192],"argmax":[195,228],"deficit,":[196],"whereas":[197],"separately":[199],"trained":[200],"discrete":[201],"learner":[202],"was":[203],"outperformed":[204],"even":[205],"by":[206],"policy,":[209],"so":[210],"deficit":[212],"not":[214],"merely":[215],"training\u2013deployment":[216],"mismatch.":[217],"Continuous":[218],"actions":[219],"remained":[220],"materially":[221],"fractional,":[222],"mean":[224],"discarded":[225,264],"mass":[226,265],"between":[229],"approximately":[230],"0.45":[231],"0.53,":[233],"but":[234],"geometry":[236],"showed":[237],"no":[238],"measurable":[239],"association":[240],"size":[243],"effect.":[247],"Argmax":[248],"therefore":[249],"removes":[250],"substantial":[251],"structure":[253],"interpretations":[258],"different":[260],"system":[261],"behavior,":[262],"without":[263],"being":[266],"identified":[267],"causal":[270],"mediator.":[271],"Deployment-aware":[272],"evaluation":[273],"should":[274],"report":[275],"both":[276],"executable":[282],"interpretation":[283],"applied":[284],"after":[285],"inference.":[286]},"counts_by_year":[],"updated_date":"2026-10-29T09:56:32.521200","created_date":"2026-10-10T00:00:00"}