{"id":"https://openalex.org/W7162430035","doi":"https://doi.org/10.48550/arxiv.2605.25123","title":"Inference-Time Alignment of Diffusion Models via Trust-Region Iterative Twisted Sequential Monte Carlo","display_name":"Inference-Time Alignment of Diffusion Models via Trust-Region Iterative Twisted Sequential Monte Carlo","publication_year":2026,"publication_date":"2026-05-24","ids":{"openalex":"https://openalex.org/W7162430035","doi":"https://doi.org/10.48550/arxiv.2605.25123"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.25123","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25123","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.2605.25123","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137077086","display_name":"Weixin Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Weixin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137070652","display_name":"Yu Chul Yang","orcid":"https://orcid.org/0000-0003-1009-4621"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136998191","display_name":"Wei Deng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Deng, Wei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137073557","display_name":"Pan Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Pan","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/T11304","display_name":"Advanced Neuroimaging Techniques and Applications","score":0.3718999922275543,"subfield":{"id":"https://openalex.org/subfields/2741","display_name":"Radiology, Nuclear Medicine and Imaging"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}},"topics":[{"id":"https://openalex.org/T11304","display_name":"Advanced Neuroimaging Techniques and Applications","score":0.3718999922275543,"subfield":{"id":"https://openalex.org/subfields/2741","display_name":"Radiology, Nuclear Medicine and Imaging"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}},{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.1662999987602234,"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/T11206","display_name":"Model Reduction and Neural Networks","score":0.06909999996423721,"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/monte-carlo-method","display_name":"Monte Carlo method","score":0.6798999905586243},{"id":"https://openalex.org/keywords/path","display_name":"Path (computing)","score":0.6406000256538391},{"id":"https://openalex.org/keywords/particle-filter","display_name":"Particle filter","score":0.619700014591217},{"id":"https://openalex.org/keywords/parameterized-complexity","display_name":"Parameterized complexity","score":0.5716999769210815},{"id":"https://openalex.org/keywords/degeneracy","display_name":"Degeneracy (biology)","score":0.5009999871253967},{"id":"https://openalex.org/keywords/base","display_name":"Base (topology)","score":0.4959000051021576},{"id":"https://openalex.org/keywords/diffusion","display_name":"Diffusion","score":0.43149998784065247},{"id":"https://openalex.org/keywords/residual","display_name":"Residual","score":0.42329999804496765},{"id":"https://openalex.org/keywords/function","display_name":"Function (biology)","score":0.42100000381469727}],"concepts":[{"id":"https://openalex.org/C19499675","wikidata":"https://www.wikidata.org/wiki/Q232207","display_name":"Monte Carlo method","level":2,"score":0.6798999905586243},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.6406000256538391},{"id":"https://openalex.org/C52421305","wikidata":"https://www.wikidata.org/wiki/Q1151499","display_name":"Particle filter","level":3,"score":0.619700014591217},{"id":"https://openalex.org/C165464430","wikidata":"https://www.wikidata.org/wiki/Q1570441","display_name":"Parameterized complexity","level":2,"score":0.5716999769210815},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5656999945640564},{"id":"https://openalex.org/C2777727622","wikidata":"https://www.wikidata.org/wiki/Q5251772","display_name":"Degeneracy (biology)","level":2,"score":0.5009999871253967},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.4959000051021576},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.47200000286102295},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.4593999981880188},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.4438000023365021},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.43149998784065247},{"id":"https://openalex.org/C155512373","wikidata":"https://www.wikidata.org/wiki/Q287450","display_name":"Residual","level":2,"score":0.42329999804496765},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.42100000381469727},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.40639999508857727},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.4041000008583069},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.3937000036239624},{"id":"https://openalex.org/C527412718","wikidata":"https://www.wikidata.org/wiki/Q855395","display_name":"Interpretation (philosophy)","level":2,"score":0.3910999894142151},{"id":"https://openalex.org/C159694833","wikidata":"https://www.wikidata.org/wiki/Q2321565","display_name":"Iterative method","level":2,"score":0.36880001425743103},{"id":"https://openalex.org/C196083921","wikidata":"https://www.wikidata.org/wiki/Q7915758","display_name":"Variance (accounting)","level":2,"score":0.35989999771118164},{"id":"https://openalex.org/C2777472644","wikidata":"https://www.wikidata.org/wiki/Q16968992","display_name":"Approximate inference","level":3,"score":0.34709998965263367},{"id":"https://openalex.org/C111350023","wikidata":"https://www.wikidata.org/wiki/Q1191869","display_name":"Markov chain Monte Carlo","level":3,"score":0.3402000069618225},{"id":"https://openalex.org/C63320529","wikidata":"https://www.wikidata.org/wiki/Q7269435","display_name":"Quasi-Monte Carlo method","level":5,"score":0.3352000117301941},{"id":"https://openalex.org/C13153151","wikidata":"https://www.wikidata.org/wiki/Q1639846","display_name":"Hybrid Monte Carlo","level":4,"score":0.3231000006198883},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.3165999948978424},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.2770000100135803},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.2720000147819519},{"id":"https://openalex.org/C121864883","wikidata":"https://www.wikidata.org/wiki/Q677916","display_name":"Statistical physics","level":1,"score":0.26820001006126404},{"id":"https://openalex.org/C2779304628","wikidata":"https://www.wikidata.org/wiki/Q3503480","display_name":"Face (sociological concept)","level":2,"score":0.262800008058548}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.25123","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25123","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.2605.25123","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25123","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":{"We":[0,128,200],"study":[1],"inference-time":[2,146,247],"alignment":[3,117,235],"for":[4,105,140],"diffusion-based":[5,116],"generative":[6],"models,":[7],"aiming":[8],"to":[9,42,76,84,115,172],"steer":[10],"a":[11,33,137,160,197,220],"base":[12,44],"model":[13],"toward":[14,210],"high-reward":[15],"outputs":[16],"without":[17],"updating":[18],"its":[19],"weights.":[20],"Recent":[21],"Sequential":[22,133],"Monte":[23,134],"Carlo":[24,135],"(SMC)-based":[25],"steering":[26],"methods":[27,60,101],"approximate":[28],"reward-tilted":[29],"target":[30,170,212],"distributions":[31],"in":[32,94,144,155],"principled":[34],"way,":[35],"but":[36],"their":[37],"proposals":[38],"remain":[39],"largely":[40],"tied":[41],"the":[43,173,184,188,203,211,215,225],"sampler.":[45,199],"Since":[46],"reward":[47],"information":[48],"is":[49,83,219],"mainly":[50,104],"used":[51],"after":[52],"propagation":[53],"through":[54],"particle":[55,64,81],"reweighting":[56],"and":[57,66,71,79,109,122,167,192,223,242],"resampling,":[58],"these":[59],"can":[61,110],"require":[62],"large":[63],"budgets":[65],"suffer":[67],"from":[68],"weight":[69],"degeneracy":[70],"high-variance":[72],"estimates.":[73],"One":[74],"way":[75],"reduce":[77],"variance":[78],"improve":[80],"efficiency":[82],"iteratively":[85],"learn":[86],"twisting":[87,100,142,190],"functions":[88,143],"that":[89,194,202,214,224],"provide":[90],"look-ahead":[91],"guidance,":[92],"as":[93],"twisted":[95,175],"SMC.":[96],"However,":[97],"existing":[98],"learnable":[99],"are":[102],"developed":[103],"classical":[106],"sequential":[107],"inference":[108],"be":[111],"unstable":[112],"when":[113],"applied":[114],"with":[118],"high-dimensional":[119],"state":[120],"spaces":[121],"terminal,":[123],"noisy,":[124],"or":[125],"black-box":[126],"rewards.":[127],"propose":[129],"Trust-Region":[130],"Iterative":[131],"Twisted":[132],"(TRI-TSMC),":[136],"trust-region":[138,204],"framework":[139],"learning":[141],"SMC-based":[145],"alignment.":[147],"Each":[148],"iteration":[149],"computes":[150],"an":[151,207],"exact":[152],"KL-constrained":[153],"update":[154,205,218],"path":[156,209,226],"space,":[157],"which":[158],"admits":[159],"closed-form":[161],"solution":[162],"by":[163,177],"tempered":[164],"importance":[165],"reweighting,":[166],"projects":[168],"this":[169],"back":[171],"parameterized":[174],"family":[176],"weighted":[178,216],"maximum":[179],"likelihood.":[180],"Theoretically,":[181],"we":[182],"formalize":[183],"value-function":[185],"interpretation":[186],"of":[187],"optimal":[189],"function":[191],"show":[193],"it":[195],"yields":[196],"zero-variance":[198],"prove":[201],"follows":[206],"escort":[208],"distribution,":[213],"maximum-likelihood":[217],"forward-KL":[221],"projection,":[222],"reduces":[227],"residual":[228],"importance-weight":[229],"variance.":[230],"Empirically,":[231],"TRI-TSMC":[232],"improves":[233],"primary":[234],"objectives":[236],"on":[237],"discrete":[238],"diffusion":[239],"text":[240],"generation":[241,244],"text-to-image":[243],"under":[245],"matched":[246],"budgets.":[248]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-27T00:00:00"}