{"id":"https://openalex.org/W7220928305","doi":"https://doi.org/10.48550/arxiv.2610.08989","title":"LASER: Latent Space Adjoint Matching for Support-Constrained Entropy-Regularized Offline RL","display_name":"LASER: Latent Space Adjoint Matching for Support-Constrained Entropy-Regularized Offline RL","publication_year":2026,"publication_date":"2026-10-06","ids":{"openalex":"https://openalex.org/W7220928305","doi":"https://doi.org/10.48550/arxiv.2610.08989"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2610.08989","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.08989","pdf_url":"https://arxiv.org/pdf/2610.08989","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":null,"license_id":null,"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/2610.08989","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5103236448","display_name":"Songyuan Zhang","orcid":"https://orcid.org/0009-0005-6465-4833","observed_orcids":["https://orcid.org/0009-0005-6465-4833"]},"institutions":[{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Zhang, Songyuan","raw_affiliation_strings":["MIT"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT","institution_ids":["https://openalex.org/I63966007"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5023512223","display_name":"Oswin So","orcid":"https://orcid.org/0000-0002-5411-3663","observed_orcids":["https://orcid.org/0000-0002-5411-3663"]},"institutions":[{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"So, Oswin","raw_affiliation_strings":["MIT"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT","institution_ids":["https://openalex.org/I63966007"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5012689023","display_name":"Eric Yang Yu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yu, Eric Yang","raw_affiliation_strings":["MIT"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT","institution_ids":["https://openalex.org/I63966007"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5035774013","display_name":"Matthew Cleaveland","orcid":"https://orcid.org/0000-0003-3671-5297","observed_orcids":["https://orcid.org/0000-0003-3671-5297"]},"institutions":[{"id":"https://openalex.org/I4210122954","display_name":"MIT Lincoln Laboratory","ror":"https://ror.org/022z6jk58","country_code":"US","type":"facility","lineage":["https://openalex.org/I4210122954","https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Cleaveland, Matthew","raw_affiliation_strings":["MIT Lincoln Laboratory"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT Lincoln Laboratory","institution_ids":["https://openalex.org/I4210122954"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125176404","display_name":"Peter Crowley-Dolen","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210122954","display_name":"MIT Lincoln Laboratory","ror":"https://ror.org/022z6jk58","country_code":"US","type":"facility","lineage":["https://openalex.org/I4210122954","https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Crowley-Dolen, Peter","raw_affiliation_strings":["MIT Lincoln Laboratory"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT Lincoln Laboratory","institution_ids":["https://openalex.org/I4210122954"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5019603699","display_name":"Chuchu Fan","orcid":"https://orcid.org/0000-0003-4671-233X","observed_orcids":["https://orcid.org/0000-0003-4671-233X"]},"institutions":[{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Fan, Chuchu","raw_affiliation_strings":["MIT"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT","institution_ids":["https://openalex.org/I63966007"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":1.0,"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":1.0,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.0,"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/T12794","display_name":"Adaptive Dynamic Programming Control","score":0.0,"subfield":{"id":"https://openalex.org/subfields/2207","display_name":"Control and Systems Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/offline-reinforcement-learning","display_name":"offline reinforcement learning","score":0.9860000014305115},{"id":"https://openalex.org/keywords/flow-matching","display_name":"flow matching","score":0.7860000133514404},{"id":"https://openalex.org/keywords/entropy-regularization","display_name":"entropy regularization","score":0.9649999737739563},{"id":"https://openalex.org/keywords/behavioral-cloning","display_name":"behavior cloning","score":0.7070000171661377},{"id":"https://openalex.org/keywords/policy-collapse","display_name":"policy collapse","score":0.6729999780654907},{"id":"https://openalex.org/keywords/backpropagation-through-time","display_name":"backpropagation through time","score":0.6949999928474426}],"concepts":[{"id":"https://openalex.org/C8642999","wikidata":"https://www.wikidata.org/wiki/Q4171168","display_name":"Hyperparameter","level":2,"score":0.7192000150680542},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6797000169754028},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6539999842643738},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.5773000121116638},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5360999703407288},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.491100013256073},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.44909998774528503},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.42260000109672546},{"id":"https://openalex.org/C106301342","wikidata":"https://www.wikidata.org/wiki/Q4117933","display_name":"Entropy (arrow of time)","level":2,"score":0.41449999809265137},{"id":"https://openalex.org/C49937458","wikidata":"https://www.wikidata.org/wiki/Q2599292","display_name":"Probabilistic logic","level":2,"score":0.4108000099658966},{"id":"https://openalex.org/C2779436431","wikidata":"https://www.wikidata.org/wiki/Q30672407","display_name":"Policy learning","level":2,"score":0.29170000553131104},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.2874000072479248},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.28349998593330383},{"id":"https://openalex.org/C2776330181","wikidata":"https://www.wikidata.org/wiki/Q18358244","display_name":"Maximization","level":2,"score":0.27379998564720154},{"id":"https://openalex.org/C51167844","wikidata":"https://www.wikidata.org/wiki/Q4422623","display_name":"Latent variable","level":2,"score":0.2689000070095062},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.26170000433921814},{"id":"https://openalex.org/C48677424","wikidata":"https://www.wikidata.org/wiki/Q6888088","display_name":"Mode (computer interface)","level":2,"score":0.25769999623298645},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2529999911785126},{"id":"https://openalex.org/C137836250","wikidata":"https://www.wikidata.org/wiki/Q984063","display_name":"Optimization problem","level":2,"score":0.25}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2610.08989","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.08989","pdf_url":"https://arxiv.org/pdf/2610.08989","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2610.08989","doi":"https://doi.org/10.48550/arxiv.2610.08989","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2610.08989","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2610.08989","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.08989","pdf_url":"https://arxiv.org/pdf/2610.08989","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G7275359387","display_name":"CAREER: DeepCertify: Data-driven Formal Approach to Safe Autonomy","funder_award_id":"2238030","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7220928305.pdf","grobid_xml":"https://content.openalex.org/works/W7220928305.grobid-xml"},"referenced_works_count":19,"referenced_works":["https://openalex.org/W2899663614","https://openalex.org/W2953981431","https://openalex.org/W2978455699","https://openalex.org/W2991355586","https://openalex.org/W2995706821","https://openalex.org/W3022566517","https://openalex.org/W3034786558","https://openalex.org/W3037932933","https://openalex.org/W3110257065","https://openalex.org/W4288099666","https://openalex.org/W4300980573","https://openalex.org/W4318242039","https://openalex.org/W4366850766","https://openalex.org/W4403662179","https://openalex.org/W4414942645","https://openalex.org/W4415057179","https://openalex.org/W4415164172","https://openalex.org/W4416611812","https://openalex.org/W7155099657"],"related_works":["https://openalex.org/W7134053873","https://openalex.org/W4309210950","https://openalex.org/W3101483449","https://openalex.org/W7160596644","https://openalex.org/W7138133154","https://openalex.org/W4309801068","https://openalex.org/W4417472120","https://openalex.org/W4391632038","https://openalex.org/W3129049380","https://openalex.org/W4380551983"],"abstract_inverted_index":{"While":[0],"offline":[1,94],"reinforcement":[2],"learning":[3,31],"(RL)":[4],"enables":[5],"policy":[6,34,52,57],"optimization":[7],"from":[8],"static":[9],"datasets":[10],"without":[11],"costly":[12],"online":[13],"interaction,":[14],"it":[15],"remains":[16],"bottlenecked":[17],"by":[18,30],"the":[19,50,56,69,146,158],"risk":[20],"of":[21,68,161],"executing":[22],"out-of-distribution":[23],"(OOD)":[24],"actions.":[25],"Recent":[26],"approaches":[27],"mitigate":[28],"this":[29,73],"a":[32,61,92],"behavior-cloning":[33],"through":[35,114],"flow":[36,109],"matching":[37,101],"and":[38,144,153],"then":[39],"performing":[40],"RL":[41,84,95,106],"within":[42],"its":[43],"constrained":[44],"latent":[45,51],"space.":[46],"However,":[47],"naively":[48],"optimizing":[49],"can":[53],"easily":[54],"cause":[55],"to":[58,102],"collapse":[59],"into":[60],"brittle":[62],"mode":[63],"or":[64],"exploit":[65],"sharp":[66],"artifacts":[67],"learned":[70],"critic.":[71],"In":[72],"work,":[74],"we":[75,128],"find":[76],"that":[77,97,130],"entropy":[78],"regularization":[79],"is":[80],"essential":[81],"in":[82],"latent-space":[83,99,105],"for":[85],"addressing":[86],"these":[87],"challenges.":[88],"We":[89],"introduce":[90],"LASER,":[91],"novel":[93],"algorithm":[96],"applies":[98],"adjoint":[100],"achieve":[103],"entropy-regularized":[104],"with":[107,124,151],"expressive":[108],"policies":[110],"while":[111],"avoiding":[112],"backpropagation":[113],"time.":[115],"Through":[116],"comprehensive":[117],"experiments":[118],"on":[119],"40":[120],"challenging":[121],"OGBench":[122],"tasks":[123,143],"varying":[125],"dataset":[126],"qualities,":[127],"show":[129],"LASER":[131,136],"achieves":[132],"state-of-the-art":[133],"performance.":[134],"Notably,":[135],"uses":[137],"fixed":[138],"method-specific":[139],"hyperparameters":[140],"across":[141],"all":[142],"outperforms":[145],"evaluated":[147],"baselines,":[148],"including":[149],"those":[150],"task-":[152],"dataset-specific":[154],"tuning,":[155],"which":[156],"highlights":[157],"robust":[159],"applicability":[160],"LASER.":[162],"Project":[163],"website:":[164],"https://mit-realm.github.io/laser/.":[165]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-10-09T00:00:00"}