{"id":"https://openalex.org/W7225757860","doi":"https://doi.org/10.48550/arxiv.2611.11821","title":"Direct Trajectory Optimization for Inference-Time Alignment of Text-to-Image Diffusion Models","display_name":"Direct Trajectory Optimization for Inference-Time Alignment of Text-to-Image Diffusion Models","publication_year":2026,"publication_date":"2026-11-09","ids":{"openalex":"https://openalex.org/W7225757860","doi":"https://doi.org/10.48550/arxiv.2611.11821"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2611.11821","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.11821","pdf_url":"https://arxiv.org/pdf/2611.11821","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2611.11821","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5029626519","display_name":"Huixuan Zhang","orcid":"https://orcid.org/0009-0009-2354-723X","observed_orcids":["https://orcid.org/0009-0009-2354-723X"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Huixuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161431001","display_name":"Li Lin","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161466033","display_name":"Junzhe Zhang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Junzhe","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5124254719","display_name":"Xiaojun Wan","orcid":"https://orcid.org/0000-0001-6887-1994","observed_orcids":["https://orcid.org/0000-0001-6887-1994"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wan, Xiaojun","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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":1.0,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":1.0,"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":9.999999747378752e-05,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.0,"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/text-to-image-diffusion-models","display_name":"text-to-image diffusion models","score":0.9869999885559082},{"id":"https://openalex.org/keywords/inference-time-alignment","display_name":"inference-time alignment","score":0.9330000281333923},{"id":"https://openalex.org/keywords/reward-modeling","display_name":"reward modeling","score":0.503000020980835},{"id":"https://openalex.org/keywords/training-free-alignment","display_name":"training-free alignment","score":0.75},{"id":"https://openalex.org/keywords/controllable-generation","display_name":"controllable generation","score":0.6309999823570251},{"id":"https://openalex.org/keywords/out-of-distribution-robustness","display_name":"out-of-distribution robustness","score":0.7039999961853027},{"id":"https://openalex.org/keywords/derivative-free-optimization","display_name":"derivative-free optimization","score":0.6899999976158142}],"concepts":[{"id":"https://openalex.org/C163294075","wikidata":"https://www.wikidata.org/wiki/Q581861","display_name":"Noise reduction","level":2,"score":0.7117999792098999},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.7005000114440918},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6410999894142151},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.59170001745224},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.5030999779701233},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5004000067710876},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.47440001368522644},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.44600000977516174},{"id":"https://openalex.org/C173246807","wikidata":"https://www.wikidata.org/wiki/Q7833062","display_name":"Trajectory optimization","level":3,"score":0.439300000667572},{"id":"https://openalex.org/C137836250","wikidata":"https://www.wikidata.org/wiki/Q984063","display_name":"Optimization problem","level":2,"score":0.43160000443458557},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.41659998893737793},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.40540000796318054},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.37929999828338623},{"id":"https://openalex.org/C2987595161","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Optimization algorithm","level":2,"score":0.3172999918460846},{"id":"https://openalex.org/C155032097","wikidata":"https://www.wikidata.org/wiki/Q798503","display_name":"Backpropagation","level":3,"score":0.310699999332428},{"id":"https://openalex.org/C47446073","wikidata":"https://www.wikidata.org/wiki/Q5165890","display_name":"Control theory (sociology)","level":3,"score":0.3086000084877014},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.28949999809265137},{"id":"https://openalex.org/C55660270","wikidata":"https://www.wikidata.org/wiki/Q5164377","display_name":"Constrained optimization","level":2,"score":0.27810001373291016},{"id":"https://openalex.org/C29265498","wikidata":"https://www.wikidata.org/wiki/Q7047719","display_name":"Noise measurement","level":3,"score":0.2727999985218048},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.2515999972820282}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2611.11821","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.11821","pdf_url":"https://arxiv.org/pdf/2611.11821","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2611.11821","doi":"https://doi.org/10.48550/arxiv.2611.11821","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.11821","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2611.11821","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.11821","pdf_url":"https://arxiv.org/pdf/2611.11821","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4403347013","https://openalex.org/W4399198301","https://openalex.org/W4414889108","https://openalex.org/W4413147029","https://openalex.org/W4387430844","https://openalex.org/W7162776841","https://openalex.org/W7127363472","https://openalex.org/W4416779202","https://openalex.org/W7155063525","https://openalex.org/W4407759100"],"abstract_inverted_index":{"Alignment":[0],"of":[1,57,112],"text-to-image":[2],"diffusion":[3,58,125,153],"models":[4,126],"with":[5],"reward":[6,113],"functions":[7],"is":[8,145],"a":[9,45,108,146],"key":[10],"challenge":[11],"in":[12],"controllable":[13],"generation.":[14],"Existing":[15],"approaches":[16],"typically":[17],"rely":[18],"on":[19,127],"optimizing":[20,53],"model":[21,101,154],"inputs":[22],"or":[23],"fine-tuning":[24],"denoising":[25,34,64,80],"networks,":[26],"which":[27,92],"requires":[28],"computationally":[29],"expensive":[30],"backpropagation":[31],"through":[32,78],"large":[33],"models.":[35,59],"In":[36],"this":[37],"work,":[38],"we":[39,87],"propose":[40],"Direct":[41],"Trajectory":[42],"Optimization":[43],"(DTO),":[44],"training-free":[46],"framework":[47],"that":[48,142],"performs":[49],"alignment":[50,121,133],"by":[51],"directly":[52],"the":[54,70,74,79,84,95,99],"sampling":[55],"trajectory":[56],"Specifically,":[60],"DTO":[61,119],"treats":[62],"intermediate":[63],"scores":[65,97],"and":[66,98,151],"sampled":[67],"noise":[68],"as":[69],"optimization":[71,144],"variables,":[72],"eliminating":[73],"need":[75],"for":[76,149],"gradients":[77],"network.":[81],"To":[82],"mitigate":[83],"out-of-distribution":[85],"problem,":[86],"introduce":[88],"Reward-Optimized":[89],"Score":[90],"Guidance,":[91],"interpolates":[93],"between":[94],"optimized":[96],"pretrained":[100],"predictions":[102],"to":[103],"stabilize":[104],"sampling,":[105],"while":[106,135],"providing":[107],"provable":[109],"theoretical":[110],"guarantee":[111],"improvement":[114],"under":[115],"constrained":[116],"distributions.":[117],"Empirically,":[118],"improves":[120],"performance":[122],"across":[123],"three":[124],"two":[128],"benchmarks,":[129],"outperforming":[130],"existing":[131],"inference-time":[132],"methods":[134],"maintaining":[136],"high":[137],"efficiency.":[138],"These":[139],"results":[140],"suggest":[141],"trajectory-level":[143],"promising":[147],"direction":[148],"efficient":[150],"effective":[152],"alignment.":[155]},"counts_by_year":[],"updated_date":"2026-11-13T13:42:25.865203","created_date":"2026-11-11T00:00:00"}