{"id":"https://openalex.org/W7161709727","doi":"https://doi.org/10.48550/arxiv.2605.16348","title":"Flow-Direct: Feedback-Efficient and Reusable Guidance for Flow Models via Non-Parametric Guidance Field","display_name":"Flow-Direct: Feedback-Efficient and Reusable Guidance for Flow Models via Non-Parametric Guidance Field","publication_year":2026,"publication_date":"2026-05-08","ids":{"openalex":"https://openalex.org/W7161709727","doi":"https://doi.org/10.48550/arxiv.2605.16348"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.16348","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.16348","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":"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.16348","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5136488571","display_name":"Kim Yong Tan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tan, Kim Yong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136487087","display_name":"Yueming Lyu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lyu, Yueming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136499359","display_name":"Ivor Tsang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tsang, Ivor","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136477252","display_name":"Yew-Soon Ong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ong, Yew-Soon","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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.09769999980926514,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.09769999980926514,"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/T10360","display_name":"Fluid Dynamics and Turbulent Flows","score":0.09380000084638596,"subfield":{"id":"https://openalex.org/subfields/2206","display_name":"Computational Mechanics"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11206","display_name":"Model Reduction and Neural Networks","score":0.09210000187158585,"subfield":{"id":"https://openalex.org/subfields/3109","display_name":"Statistical and Nonlinear Physics"},"field":{"id":"https://openalex.org/fields/31","display_name":"Physics and Astronomy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/field","display_name":"Field (mathematics)","score":0.6952999830245972},{"id":"https://openalex.org/keywords/estimator","display_name":"Estimator","score":0.635200023651123},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.5968999862670898},{"id":"https://openalex.org/keywords/sample","display_name":"Sample (material)","score":0.546999990940094},{"id":"https://openalex.org/keywords/flow","display_name":"Flow (mathematics)","score":0.400299996137619},{"id":"https://openalex.org/keywords/guidance-system","display_name":"Guidance system","score":0.39399999380111694}],"concepts":[{"id":"https://openalex.org/C9652623","wikidata":"https://www.wikidata.org/wiki/Q190109","display_name":"Field (mathematics)","level":2,"score":0.6952999830245972},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.635200023651123},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6208000183105469},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.5968999862670898},{"id":"https://openalex.org/C198531522","wikidata":"https://www.wikidata.org/wiki/Q485146","display_name":"Sample (material)","level":2,"score":0.546999990940094},{"id":"https://openalex.org/C38349280","wikidata":"https://www.wikidata.org/wiki/Q1434290","display_name":"Flow (mathematics)","level":2,"score":0.400299996137619},{"id":"https://openalex.org/C201004817","wikidata":"https://www.wikidata.org/wiki/Q1707071","display_name":"Guidance system","level":2,"score":0.39399999380111694},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.376800000667572},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.3395000100135803},{"id":"https://openalex.org/C133731056","wikidata":"https://www.wikidata.org/wiki/Q4917288","display_name":"Control engineering","level":1,"score":0.32820001244544983},{"id":"https://openalex.org/C47446073","wikidata":"https://www.wikidata.org/wiki/Q5165890","display_name":"Control theory (sociology)","level":3,"score":0.3025999963283539},{"id":"https://openalex.org/C2779136372","wikidata":"https://www.wikidata.org/wiki/Q10283002","display_name":"Information flow","level":2,"score":0.29789999127388},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.26080000400543213},{"id":"https://openalex.org/C3020318244","wikidata":"https://www.wikidata.org/wiki/Q4812187","display_name":"Large sample","level":2,"score":0.2574000060558319},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.2565999925136566}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.16348","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.16348","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.16348","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.16348","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":"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"},"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":{"Training-free":[0],"guidance":[1,63,67,117,144,167,180],"enables":[2],"pre-trained":[3,86],"diffusion":[4],"and":[5,42,79,178],"flow":[6],"models":[7],"to":[8,31,88,140,185],"optimize":[9],"application-specific":[10],"objectives":[11],"using":[12],"feedback":[13,26],"from":[14,72,103],"external":[15],"black-box":[16],"reward":[17,25,147,176],"functions.":[18],"However,":[19],"existing":[20],"methods":[21],"are":[22,111],"feedback-inefficient":[23],"because":[24,134],"is":[27,43,69,96,131,138,149,154,159],"used":[28,139],"only":[29],"transiently":[30],"inform":[32],"a":[33,38,53,61,99,165],"localized":[34],"gradient":[35],"approximation":[36],"or":[37],"discrete":[39],"search":[40],"decision,":[41],"subsequently":[44],"discarded.":[45],"To":[46],"address":[47],"this":[48,66,115],"limitation,":[49],"we":[50],"propose":[51],"Flow-Direct,":[52],"framework":[54,153],"that":[55,188],"guides":[56],"the":[57,73,77,85,89,94,142,152,161],"generation":[58],"process":[59],"via":[60],"persistent":[62,123],"field.":[64],"Theoretically,":[65],"field":[68,95,118,168],"analytically":[70],"derived":[71],"log-density":[74],"ratio":[75],"between":[76],"base":[78],"reward-weighted":[80],"target":[81,90,172],"distributions;":[82],"it":[83],"transports":[84],"distribution":[87],"distribution.":[91],"In":[92],"practice,":[93],"implemented":[97],"as":[98],"non-parametric":[100],"estimator":[101],"constructed":[102],"all":[104],"accumulated":[105],"reward-evaluated":[106],"samples.":[107],"As":[108],"more":[109],"samples":[110,173,187],"collected":[112,162],"during":[113],"optimization,":[114],"empirical":[116],"becomes":[119],"increasingly":[120],"accurate.":[121],"This":[122],"formulation":[124],"yields":[125],"two":[126],"major":[127],"advantages.":[128],"First,":[129],"Flow-Direct":[130],"highly":[132],"feedback-efficient:":[133],"every":[135],"evaluated":[136],"sample":[137],"refine":[141],"global":[143],"field,":[145],"no":[146],"information":[148],"wasted.":[150],"Second,":[151],"naturally":[155],"reusable:":[156],"once":[157],"optimization":[158],"complete,":[160],"dataset":[163],"defines":[164],"reusable":[166],"for":[169],"generating":[170],"novel":[171],"without":[174],"additional":[175],"evaluations,":[177],"distinct":[179],"fields":[181],"can":[182],"be":[183],"combined":[184],"generate":[186],"simultaneously":[189],"satisfy":[190],"multiple":[191],"objectives.":[192]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-20T00:00:00"}