{"id":"https://openalex.org/W7214965162","doi":"https://doi.org/10.48550/arxiv.2609.37441","title":"Anisotropic Representations Improve Planning in JEPA World Models","display_name":"Anisotropic Representations Improve Planning in JEPA World Models","publication_year":2026,"publication_date":"2026-09-29","ids":{"openalex":"https://openalex.org/W7214965162","doi":"https://doi.org/10.48550/arxiv.2609.37441"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.37441","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37441","pdf_url":"https://arxiv.org/pdf/2609.37441","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/2609.37441","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5046567039","display_name":"Kang Mingu","orcid":"https://orcid.org/0009-0004-8437-7629","observed_orcids":["https://orcid.org/0009-0004-8437-7629"]},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Kang, Mingu","raw_affiliation_strings":["Seoul National University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Seoul National University","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5020776717","display_name":"Yoori Oh","orcid":"https://orcid.org/0009-0002-2557-0934","observed_orcids":["https://orcid.org/0009-0002-2557-0934"]},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Oh, Yoori","raw_affiliation_strings":["Seoul National University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Seoul National University","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5074194612","display_name":"Sookyung Kim","orcid":"https://orcid.org/0000-0001-6364-6907","observed_orcids":["https://orcid.org/0000-0001-6364-6907","https://orcid.org/0000-0002-6413-9920"]},"institutions":[{"id":"https://openalex.org/I138925566","display_name":"Ewha Womans University","ror":"https://ror.org/053fp5c05","country_code":"KR","type":"education","lineage":["https://openalex.org/I138925566"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Kim, Sookyung","raw_affiliation_strings":["Ewha Womans University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Ewha Womans University","institution_ids":["https://openalex.org/I138925566"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5153678135","display_name":"Joonseok Lee","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Lee, Joonseok","raw_affiliation_strings":["Seoul National University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Seoul National University","institution_ids":["https://openalex.org/I139264467"]}]}],"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":0.9970999956130981,"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":0.9970999956130981,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.0015999999595806003,"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":0.00019999999494757503,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/latent-world-models","display_name":"latent world models","score":0.8420000076293945},{"id":"https://openalex.org/keywords/jepa","display_name":"JEPA","score":0.8220000267028809},{"id":"https://openalex.org/keywords/anisotropic-regularization","display_name":"anisotropic regularization","score":0.7429999709129333},{"id":"https://openalex.org/keywords/diagonal-covariance","display_name":"diagonal covariance","score":0.5350000262260437},{"id":"https://openalex.org/keywords/representation-collapse","display_name":"representation collapse","score":0.7039999961853027},{"id":"https://openalex.org/keywords/representational-geometry","display_name":"representational geometry","score":0.609000027179718},{"id":"https://openalex.org/keywords/planning-costs","display_name":"planning costs","score":0.4560000002384186},{"id":"https://openalex.org/keywords/visual-control","display_name":"visual control","score":0.45899999141693115}],"concepts":[{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.6219000220298767},{"id":"https://openalex.org/C163716315","wikidata":"https://www.wikidata.org/wiki/Q901177","display_name":"Gaussian","level":2,"score":0.5421000123023987},{"id":"https://openalex.org/C178650346","wikidata":"https://www.wikidata.org/wiki/Q201984","display_name":"Covariance","level":2,"score":0.5175999999046326},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.5049999952316284},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.4706999957561493},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4580000042915344},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.45260000228881836},{"id":"https://openalex.org/C130367717","wikidata":"https://www.wikidata.org/wiki/Q189791","display_name":"Diagonal","level":2,"score":0.44679999351501465},{"id":"https://openalex.org/C61326573","wikidata":"https://www.wikidata.org/wiki/Q1496376","display_name":"Gaussian process","level":3,"score":0.40529999136924744},{"id":"https://openalex.org/C129782007","wikidata":"https://www.wikidata.org/wiki/Q162886","display_name":"Euclidean geometry","level":2,"score":0.40450000762939453},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.39250001311302185},{"id":"https://openalex.org/C184050105","wikidata":"https://www.wikidata.org/wiki/Q273163","display_name":"Isotropy","level":2,"score":0.38609999418258667},{"id":"https://openalex.org/C81074085","wikidata":"https://www.wikidata.org/wiki/Q366872","display_name":"Motion planning","level":3,"score":0.38510000705718994},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.3684999942779541},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3555000126361847},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.3522999882698059},{"id":"https://openalex.org/C186450821","wikidata":"https://www.wikidata.org/wiki/Q17295","display_name":"Euclidean space","level":2,"score":0.3391000032424927},{"id":"https://openalex.org/C190470478","wikidata":"https://www.wikidata.org/wiki/Q2370229","display_name":"Invariant (physics)","level":2,"score":0.32269999384880066},{"id":"https://openalex.org/C120174047","wikidata":"https://www.wikidata.org/wiki/Q847073","display_name":"Euclidean distance","level":2,"score":0.3147999942302704},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.30970001220703125},{"id":"https://openalex.org/C185142706","wikidata":"https://www.wikidata.org/wiki/Q1134404","display_name":"Covariance matrix","level":2,"score":0.29589998722076416},{"id":"https://openalex.org/C31170391","wikidata":"https://www.wikidata.org/wiki/Q188619","display_name":"Hierarchy","level":2,"score":0.2896000146865845},{"id":"https://openalex.org/C70437156","wikidata":"https://www.wikidata.org/wiki/Q7228652","display_name":"Pooling","level":2,"score":0.27709999680519104},{"id":"https://openalex.org/C191795146","wikidata":"https://www.wikidata.org/wiki/Q3878446","display_name":"Norm (philosophy)","level":2,"score":0.27160000801086426},{"id":"https://openalex.org/C198082294","wikidata":"https://www.wikidata.org/wiki/Q3399648","display_name":"Position (finance)","level":2,"score":0.2685999870300293},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.2685000002384186},{"id":"https://openalex.org/C132459708","wikidata":"https://www.wikidata.org/wiki/Q744069","display_name":"Extrapolation","level":2,"score":0.25769999623298645}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.37441","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37441","pdf_url":"https://arxiv.org/pdf/2609.37441","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.2609.37441","doi":"https://doi.org/10.48550/arxiv.2609.37441","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.37441","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:2609.37441","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37441","pdf_url":"https://arxiv.org/pdf/2609.37441","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":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7214965162.pdf","grobid_xml":"https://content.openalex.org/works/W7214965162.grobid-xml"},"referenced_works_count":15,"referenced_works":["https://openalex.org/W4416338927","https://openalex.org/W4417261359","https://openalex.org/W7117123094","https://openalex.org/W7127358364","https://openalex.org/W7129589260","https://openalex.org/W7135196883","https://openalex.org/W7140141371","https://openalex.org/W7160855100","https://openalex.org/W7162085595","https://openalex.org/W7162566815","https://openalex.org/W7166107790","https://openalex.org/W7171812240","https://openalex.org/W7203687260","https://openalex.org/W7203913879","https://openalex.org/W7204197618"],"related_works":["https://openalex.org/W7172432896","https://openalex.org/W7214905115","https://openalex.org/W7203687260","https://openalex.org/W7160937232","https://openalex.org/W7223129368","https://openalex.org/W7204197618","https://openalex.org/W7135201772","https://openalex.org/W7171663849","https://openalex.org/W7214719230","https://openalex.org/W7202289612"],"abstract_inverted_index":{"Latent":[0],"world":[1],"models":[2],"learn":[3],"action-conditioned":[4],"dynamics":[5],"in":[6,156],"representation":[7,26,33],"space":[8],"and":[9,49,99,107,134],"often":[10],"score":[11],"candidate":[12],"actions":[13],"by":[14],"Euclidean":[15,108],"distance":[16],"to":[17,27],"a":[18,55,65,93],"goal":[19],"representation.":[20],"Joint":[21],"training":[22,132],"typically":[23],"regularizes":[24],"the":[25,31,73,87,112,122,131,135,139],"prevent":[28],"collapse,":[29],"but":[30],"resulting":[32],"geometry":[34,66],"also":[35,163],"determines":[36],"how":[37],"terminal":[38],"errors":[39],"are":[40],"weighted":[41],"during":[42,117],"planning.":[43],"We":[44],"show":[45],"that":[46,67],"accurate":[47],"prediction":[48,103],"noncollapsed":[50],"representations":[51],"do":[52],"not":[53],"guarantee":[54],"task-aligned":[56],"latent":[57,160],"planning":[58,143,152,161],"cost:":[59],"isotropic":[60,89],"Gaussian":[61,90],"regularization":[62],"can":[63],"induce":[64],"ranks":[68],"feasible":[69],"outcomes":[70],"differently":[71],"from":[72],"task":[74,168],"cost.":[75],"To":[76],"address":[77],"this":[78],"mismatch,":[79],"we":[80],"introduce":[81],"AnisoWM":[82,150],"with":[83,92,167],"$\u039b$Reg,":[84],"which":[85,138],"replaces":[86],"fixed":[88],"target":[91,113,126],"learnable":[94],"diagonal":[95],"covariance":[96],"under":[97,137],"fixed-trace":[98],"anisotropy":[100],"constraints.":[101],"The":[102],"objective,":[104],"predictor":[105],"architecture,":[106],"planner":[109],"remain":[110],"unchanged;":[111],"is":[114],"used":[115],"only":[116],"training.":[118],"Our":[119],"analysis":[120],"characterizes":[121],"prediction-driven":[123],"allocation":[124],"of":[125],"variance,":[127],"its":[128],"dependence":[129],"on":[130],"distribution,":[133],"conditions":[136],"induced":[140],"metric":[141],"reduces":[142],"regret.":[144],"Across":[145],"four":[146],"visual":[147],"control":[148],"environments,":[149],"improves":[151],"success":[153],"over":[154],"LeWorldModel":[155],"all":[157],"four.":[158],"Its":[159],"cost":[162],"shows":[164],"better":[165],"agreement":[166],"outcomes.":[169],"Project":[170],"website:":[171],"https://rkdrn79.github.io/AnisoWM-page/":[172]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-10-01T00:00:00"}