{"id":"https://openalex.org/W7214624183","doi":"https://doi.org/10.48550/arxiv.2609.31458","title":"Nonparametric In-Context Learning under Growing Geometric Complexity: Minimax Optimality and Local Geometry-Adaptivity of Transformers","display_name":"Nonparametric In-Context Learning under Growing Geometric Complexity: Minimax Optimality and Local Geometry-Adaptivity of Transformers","publication_year":2026,"publication_date":"2026-09-25","ids":{"openalex":"https://openalex.org/W7214624183","doi":"https://doi.org/10.48550/arxiv.2609.31458"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.31458","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.31458","pdf_url":"https://arxiv.org/pdf/2609.31458","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","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.31458","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5150649167","display_name":"Jaehee Seo","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":"Seo, Jaehee","raw_affiliation_strings":["Department of Statistics Seoul National University , Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Statistics Seoul National University , Korea","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5153231259","display_name":"Jisu Kim","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":"Kim, Jisu","raw_affiliation_strings":["Department of Statistics Seoul National University , Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Statistics Seoul National University , Korea","institution_ids":["https://openalex.org/I139264467"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I139264467"],"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/T12072","display_name":"Machine Learning and Algorithms","score":0.3571999967098236,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.3571999967098236,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.20399999618530273,"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/T10320","display_name":"Neural Networks and Applications","score":0.1680999994277954,"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/in-context-learning","display_name":"in-context learning","score":0.9990000128746033},{"id":"https://openalex.org/keywords/transformer","display_name":"transformer models","score":0.8799999952316284},{"id":"https://openalex.org/keywords/nonparametric-learning","display_name":"nonparametric learning","score":0.6449999809265137},{"id":"https://openalex.org/keywords/minimax-lower-bounds","display_name":"minimax lower bounds","score":0.7760000228881836},{"id":"https://openalex.org/keywords/empirical-risk-minimization","display_name":"empirical risk minimization","score":0.5809999704360962}],"concepts":[{"id":"https://openalex.org/C149728462","wikidata":"https://www.wikidata.org/wiki/Q751319","display_name":"Minimax","level":2,"score":0.7404000163078308},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.6323000192642212},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.6248000264167786},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.4986000061035156},{"id":"https://openalex.org/C102366305","wikidata":"https://www.wikidata.org/wiki/Q1097688","display_name":"Nonparametric statistics","level":2,"score":0.45570001006126404},{"id":"https://openalex.org/C77553402","wikidata":"https://www.wikidata.org/wiki/Q13222579","display_name":"Upper and lower bounds","level":2,"score":0.44929999113082886},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.4302999973297119},{"id":"https://openalex.org/C39927690","wikidata":"https://www.wikidata.org/wiki/Q11197","display_name":"Logarithm","level":2,"score":0.40950000286102295},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.38350000977516174},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.3815999925136566},{"id":"https://openalex.org/C138187205","wikidata":"https://www.wikidata.org/wiki/Q131251","display_name":"Tangent","level":2,"score":0.3531000018119812},{"id":"https://openalex.org/C133939421","wikidata":"https://www.wikidata.org/wiki/Q6865379","display_name":"Minimax estimator","level":4,"score":0.34860000014305115},{"id":"https://openalex.org/C117251300","wikidata":"https://www.wikidata.org/wiki/Q1849855","display_name":"Parametric statistics","level":2,"score":0.3393999934196472},{"id":"https://openalex.org/C22324862","wikidata":"https://www.wikidata.org/wiki/Q652707","display_name":"Lipschitz continuity","level":2,"score":0.3199000060558319},{"id":"https://openalex.org/C129782007","wikidata":"https://www.wikidata.org/wiki/Q162886","display_name":"Euclidean geometry","level":2,"score":0.29499998688697815},{"id":"https://openalex.org/C120174047","wikidata":"https://www.wikidata.org/wiki/Q847073","display_name":"Euclidean distance","level":2,"score":0.2921999990940094},{"id":"https://openalex.org/C109546454","wikidata":"https://www.wikidata.org/wiki/Q3798604","display_name":"Information geometry","level":4,"score":0.27869999408721924},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2782999873161316},{"id":"https://openalex.org/C107321475","wikidata":"https://www.wikidata.org/wiki/Q5374254","display_name":"Empirical risk minimization","level":2,"score":0.27630001306533813},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.27239999175071716},{"id":"https://openalex.org/C90119067","wikidata":"https://www.wikidata.org/wiki/Q43260","display_name":"Polynomial","level":2,"score":0.27140000462532043},{"id":"https://openalex.org/C106301342","wikidata":"https://www.wikidata.org/wiki/Q4117933","display_name":"Entropy (arrow of time)","level":2,"score":0.2655999958515167},{"id":"https://openalex.org/C165700671","wikidata":"https://www.wikidata.org/wiki/Q203484","display_name":"Laplace operator","level":2,"score":0.2547000050544739}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.31458","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.31458","pdf_url":"https://arxiv.org/pdf/2609.31458","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.2609.31458","doi":"https://doi.org/10.48550/arxiv.2609.31458","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.31458","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:2609.31458","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.31458","pdf_url":"https://arxiv.org/pdf/2609.31458","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":[{"id":"https://openalex.org/F4320321292","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7214624183.pdf","grobid_xml":"https://content.openalex.org/works/W7214624183.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4415795960","https://openalex.org/W4417357162","https://openalex.org/W7127311704","https://openalex.org/W7164542711","https://openalex.org/W4415796768","https://openalex.org/W7130510275","https://openalex.org/W7214289539","https://openalex.org/W7137897835","https://openalex.org/W4403995583","https://openalex.org/W4412122873"],"abstract_inverted_index":{"Transformers":[0],"have":[1],"become":[2],"a":[3,81,100,109,115],"central":[4],"architecture":[5],"for":[6,147],"in-context":[7,144],"learning":[8],"(ICL),":[9],"particularly":[10],"through":[11],"their":[12],"state-of-the-art":[13],"performance":[14],"in":[15,28],"large":[16],"language":[17],"models.":[18,45],"This":[19,104],"success":[20],"motivates":[21],"understanding":[22],"how":[23],"transformers":[24],"exploit":[25],"task-relevant":[26],"structure":[27],"geometrically":[29],"heterogeneous":[30,67],"data.":[31],"However,":[32],"existing":[33],"nonparametric":[34],"ICL":[35],"theory":[36],"has":[37],"largely":[38],"focused":[39],"on":[40],"Euclidean":[41],"domains":[42],"or":[43],"single-manifold":[44],"To":[46],"address":[47],"this":[48,153],"gap,":[49],"we":[50,79,141],"study":[51],"the":[52,86,90,131,162,170],"prediction":[53],"problem":[54],"under":[55,160],"unknown":[56],"local":[57,74,166],"geometry,":[58],"modeled":[59],"by":[60],"sample":[61],"size-dependent":[62],"mixtures":[63],"of":[64,89],"manifolds":[65],"with":[66,99,114,134],"dimensions,":[68],"smoothness,":[69],"and":[70,76,92,118,137,168],"sampling":[71],"masses.":[72],"Under":[73],"separation":[75],"small-perturbation":[77],"conditions,":[78],"establish":[80],"minimax":[82,132,172],"lower":[83],"bound":[84,146],"capturing":[85],"aggregate":[87,171],"difficulty":[88],"components":[91],"construct":[93],"an":[94,143],"oracle":[95],"tangent":[96],"local-polynomial":[97,121],"estimator":[98,105],"matching":[101],"upper":[102],"bound.":[103],"is":[106],"connected":[107],"to":[108,130],"structure-informed,":[110],"two-stage":[111],"softmax":[112],"transformer":[113,124],"geometric":[116],"preconditioner":[117],"chartwise":[119],"reduced":[120],"solvers.":[122],"The":[123],"achieves":[125],"negligible":[126],"approximation":[127],"error":[128],"relative":[129],"rate":[133],"logarithmic":[135],"depth":[136],"polynomial":[138],"size.":[139],"Finally,":[140],"derive":[142],"generalization":[145],"near":[148],"empirical":[149],"risk":[150],"minimizers":[151],"over":[152],"class.":[154],"Together,":[155],"these":[156],"results":[157],"identify":[158],"conditions":[159],"which":[161],"resulting":[163],"predictor":[164],"exploits":[165],"geometry":[167],"attains":[169],"rate.":[173]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-09-29T00:00:00"}