{"id":"https://openalex.org/W7160949841","doi":"https://doi.org/10.48550/arxiv.2605.08464","title":"The Geometric Structure of Models Learning Sparse Data","display_name":"The Geometric Structure of Models Learning Sparse Data","publication_year":2026,"publication_date":"2026-05-08","ids":{"openalex":"https://openalex.org/W7160949841","doi":"https://doi.org/10.48550/arxiv.2605.08464"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.08464","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.08464","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":"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.08464","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5133676478","display_name":"Thomas Walker","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Walker, Thomas","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5017342958","display_name":"T. Mitchell Roddenberry","orcid":"https://orcid.org/0000-0001-9031-6305"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Roddenberry, T. Mitchell","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135964947","display_name":"Ahmed Imtiaz Humayun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Humayun, Ahmed Imtiaz","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135951254","display_name":"Randall Balestriero","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Balestriero, Randall","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135910718","display_name":"Richard Baraniuk","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Baraniuk, Richard","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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.881600022315979,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.881600022315979,"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.04320000112056732,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.01889999955892563,"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/jacobian-matrix-and-determinant","display_name":"Jacobian matrix and determinant","score":0.6498000025749207},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.602400004863739},{"id":"https://openalex.org/keywords/curse-of-dimensionality","display_name":"Curse of dimensionality","score":0.5601000189781189},{"id":"https://openalex.org/keywords/outlier","display_name":"Outlier","score":0.5056999921798706},{"id":"https://openalex.org/keywords/regularization","display_name":"Regularization (linguistics)","score":0.4690000116825104},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.41760000586509705},{"id":"https://openalex.org/keywords/centroid","display_name":"Centroid","score":0.37770000100135803},{"id":"https://openalex.org/keywords/regular-polygon","display_name":"Regular polygon","score":0.33640000224113464},{"id":"https://openalex.org/keywords/matrix-norm","display_name":"Matrix norm","score":0.3273000121116638}],"concepts":[{"id":"https://openalex.org/C200331156","wikidata":"https://www.wikidata.org/wiki/Q506041","display_name":"Jacobian matrix and determinant","level":2,"score":0.6498000025749207},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.602400004863739},{"id":"https://openalex.org/C111030470","wikidata":"https://www.wikidata.org/wiki/Q1430460","display_name":"Curse of dimensionality","level":2,"score":0.5601000189781189},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5404999852180481},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5113000273704529},{"id":"https://openalex.org/C79337645","wikidata":"https://www.wikidata.org/wiki/Q779824","display_name":"Outlier","level":2,"score":0.5056999921798706},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.4690000116825104},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4496000111103058},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.41760000586509705},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.39089998602867126},{"id":"https://openalex.org/C146599234","wikidata":"https://www.wikidata.org/wiki/Q511093","display_name":"Centroid","level":2,"score":0.37770000100135803},{"id":"https://openalex.org/C112680207","wikidata":"https://www.wikidata.org/wiki/Q714886","display_name":"Regular polygon","level":2,"score":0.33640000224113464},{"id":"https://openalex.org/C92207270","wikidata":"https://www.wikidata.org/wiki/Q939253","display_name":"Matrix norm","level":3,"score":0.3273000121116638},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.314300000667572},{"id":"https://openalex.org/C56372850","wikidata":"https://www.wikidata.org/wiki/Q1050404","display_name":"Sparse matrix","level":3,"score":0.31150001287460327},{"id":"https://openalex.org/C529865628","wikidata":"https://www.wikidata.org/wiki/Q1790740","display_name":"Manifold (fluid mechanics)","level":2,"score":0.30630001425743103},{"id":"https://openalex.org/C75553542","wikidata":"https://www.wikidata.org/wiki/Q178161","display_name":"A priori and a posteriori","level":2,"score":0.30570000410079956},{"id":"https://openalex.org/C42812","wikidata":"https://www.wikidata.org/wiki/Q1082910","display_name":"Partition (number theory)","level":2,"score":0.30469998717308044},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.30169999599456787},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.2928999960422516},{"id":"https://openalex.org/C21080849","wikidata":"https://www.wikidata.org/wiki/Q13611879","display_name":"Data point","level":2,"score":0.28060001134872437},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.2766000032424927},{"id":"https://openalex.org/C162319229","wikidata":"https://www.wikidata.org/wiki/Q175263","display_name":"Data structure","level":2,"score":0.2727999985218048},{"id":"https://openalex.org/C2781170535","wikidata":"https://www.wikidata.org/wiki/Q30587856","display_name":"Noisy data","level":2,"score":0.2678000032901764},{"id":"https://openalex.org/C191795146","wikidata":"https://www.wikidata.org/wiki/Q3878446","display_name":"Norm (philosophy)","level":2,"score":0.2662999927997589},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.2614000141620636},{"id":"https://openalex.org/C2986090443","wikidata":"https://www.wikidata.org/wiki/Q77870413","display_name":"Local structure","level":2,"score":0.25850000977516174},{"id":"https://openalex.org/C95623464","wikidata":"https://www.wikidata.org/wiki/Q1096149","display_name":"Classifier (UML)","level":2,"score":0.2547999918460846}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.08464","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.08464","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.08464","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.08464","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":"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"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"manifold":[1,46],"hypothesis":[2],"(MH)":[3],"is":[4,21,47,57],"often":[5],"used":[6],"to":[7,178,190,195,211],"explain":[8],"how":[9],"machine":[10],"learning":[11],"can":[12],"overcome":[13],"the":[14,19,27,36,52,55,71,103,108,137,146,171,179,185],"curse":[15],"of":[16,35,174,187],"dimensionality.":[17],"However,":[18],"MH":[20,56],"only":[22],"applicable":[23,59],"in":[24,70],"regimes":[25,53],"where":[26,42,54],"training":[28,104,109,172],"data":[29,39,105],"provides":[30],"a":[31,44,76,81,120,157],"sufficiently":[32],"dense":[33],"sample":[34],"underlying":[37],"low-dimensional":[38,45],"manifold,":[40],"or":[41],"such":[43],"conceivably":[48],"present.":[49],"We":[50,88,165,202],"describe":[51],"not":[58],"as":[60,85,133],"sparse.":[61],"In":[62],"this":[63],"paper,":[64],"we":[65,83,154,183],"demonstrate":[66,166],"that":[67,90,160,167,204],"models":[68],"succeed":[69],"sparse":[72],"regime":[73],"by":[74,150],"exploiting":[75],"highly":[77],"structured":[78],"local":[79,117],"geometry,":[80],"property":[82],"formalize":[84],"normal":[86,129,163,188],"alignment.":[87,164],"prove":[89],"normal-aligned":[91],"classifiers":[92],"--":[93,106],"whose":[94],"input-output":[95],"Jacobians":[96],"are":[97],"rank-one":[98],"and":[99,114,143],"align":[100],"perfectly":[101],"with":[102],"minimize":[107],"objective":[110],"under":[111,119],"norm":[112],"constraints":[113],"achieve":[115],"maximal":[116],"robustness":[118,209],"non-zero":[121],"Jacobian":[122],"constraint.":[123],"For":[124],"continuous":[125],"piecewise-affine":[126],"deep":[127,175],"networks,":[128],"alignment":[130,135,189],"manifests":[131],"geometrically":[132],"centroid":[134],"within":[136],"network's":[138],"induced":[139],"power":[140],"diagram":[141],"partition":[142],"results":[144],"from":[145],"feature-learning":[147],"regime.":[148],"Motivated":[149],"these":[151],"theoretical":[152],"insights,":[153],"introduce":[155,196],"GrokAlign,":[156],"regularization":[158],"strategy":[159],"actively":[161],"induces":[162],"GrokAlign":[168],"significantly":[169],"accelerates":[170],"dynamics":[173],"networks":[176],"relevant":[177],"grokking":[180],"phenomenon.":[181],"Furthermore,":[182],"apply":[184],"principle":[186],"Recursive":[191,197],"Feature":[192,198],"Machines":[193,200],"(RFMs)":[194],"Alignment":[199],"(RFAMs).":[201],"show":[203],"RFAMs":[205],"exhibit":[206],"greater":[207],"adversarial":[208],"compared":[210],"RFMs":[212],"when":[213],"trained":[214],"on":[215],"tabular":[216],"data.":[217]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-13T00:00:00"}