{"id":"https://openalex.org/W7220908583","doi":"https://doi.org/10.48550/arxiv.2610.08420","title":"Symmetry-Aware Feature Learning: A Polynomial Separation for Multi-Index Models","display_name":"Symmetry-Aware Feature Learning: A Polynomial Separation for Multi-Index Models","publication_year":2026,"publication_date":"2026-10-06","ids":{"openalex":"https://openalex.org/W7220908583","doi":"https://doi.org/10.48550/arxiv.2610.08420"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2610.08420","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.08420","pdf_url":"https://arxiv.org/pdf/2610.08420","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/2610.08420","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100027159","display_name":"Jivan Waber","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Waber, Jivan","raw_affiliation_strings":["Information , Learning and Physics Laboratory"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Information , Learning and Physics Laboratory","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5106491657","display_name":"Vanessa Piccolo","orcid":"https://orcid.org/0000-0003-0792-9798","observed_orcids":["https://orcid.org/0000-0003-0792-9798"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Piccolo, Vanessa","raw_affiliation_strings":["Information , Learning and Physics Laboratory"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Information , Learning and Physics Laboratory","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039205039","display_name":"Yatin Dandi","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I5124864","display_name":"\u00c9cole Polytechnique F\u00e9d\u00e9rale de Lausanne","ror":"https://ror.org/02s376052","country_code":"CH","type":"education","lineage":["https://openalex.org/I2799323385","https://openalex.org/I5124864"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Dandi, Yatin","raw_affiliation_strings":["Information , Learning and Physics Laboratory","Statistical Physics of Computation Laboratory \u00c9cole Polytechnique F\u00e9d\u00e9rale de Lausanne (EPFL)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Information , Learning and Physics Laboratory","institution_ids":[]},{"raw_affiliation_string":"Statistical Physics of Computation Laboratory \u00c9cole Polytechnique F\u00e9d\u00e9rale de Lausanne (EPFL)","institution_ids":["https://openalex.org/I5124864"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5068236230","display_name":"Florent Krz\u0105ka\u0142a","orcid":"https://orcid.org/0000-0003-2313-2578","observed_orcids":["https://orcid.org/0000-0003-2313-2578"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Krzakala, Florent","raw_affiliation_strings":["Information , Learning and Physics Laboratory"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Information , Learning and Physics Laboratory","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":1,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.8300999999046326,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.8300999999046326,"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":0.07689999788999557,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.03929999843239784,"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/sample-complexity","display_name":"sample complexity","score":0.5720000267028809},{"id":"https://openalex.org/keywords/multi-index-models","display_name":"multi-index models","score":0.9129999876022339},{"id":"https://openalex.org/keywords/data-augmentation","display_name":"data augmentation","score":0.6940000057220459},{"id":"https://openalex.org/keywords/weight-sharing","display_name":"parameter sharing","score":0.7979999780654907},{"id":"https://openalex.org/keywords/correlation-loss","display_name":"correlation loss","score":0.7139999866485596},{"id":"https://openalex.org/keywords/weak-recovery","display_name":"weak recovery","score":0.6880000233650208}],"concepts":[{"id":"https://openalex.org/C114466953","wikidata":"https://www.wikidata.org/wiki/Q6034165","display_name":"Initialization","level":2,"score":0.7182000279426575},{"id":"https://openalex.org/C90119067","wikidata":"https://www.wikidata.org/wiki/Q43260","display_name":"Polynomial","level":2,"score":0.6162999868392944},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.48249998688697815},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.4814000129699707},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.4797999858856201},{"id":"https://openalex.org/C39927690","wikidata":"https://www.wikidata.org/wiki/Q11197","display_name":"Logarithm","level":2,"score":0.45969998836517334},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.4496999979019165},{"id":"https://openalex.org/C163716315","wikidata":"https://www.wikidata.org/wiki/Q901177","display_name":"Gaussian","level":2,"score":0.44859999418258667},{"id":"https://openalex.org/C129844170","wikidata":"https://www.wikidata.org/wiki/Q41299","display_name":"Quadratic equation","level":2,"score":0.4343999922275543},{"id":"https://openalex.org/C198531522","wikidata":"https://www.wikidata.org/wiki/Q485146","display_name":"Sample (material)","level":2,"score":0.4097999930381775},{"id":"https://openalex.org/C311688","wikidata":"https://www.wikidata.org/wiki/Q2393193","display_name":"Time complexity","level":2,"score":0.3946000039577484},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.39169999957084656},{"id":"https://openalex.org/C2780388253","wikidata":"https://www.wikidata.org/wiki/Q5421508","display_name":"Exponent","level":2,"score":0.3840000033378601},{"id":"https://openalex.org/C2779886137","wikidata":"https://www.wikidata.org/wiki/Q21030012","display_name":"Symmetry (geometry)","level":2,"score":0.35929998755455017},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3479999899864197},{"id":"https://openalex.org/C120068334","wikidata":"https://www.wikidata.org/wiki/Q45343","display_name":"Polynomial regression","level":3,"score":0.34290000796318054},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.34049999713897705},{"id":"https://openalex.org/C61326573","wikidata":"https://www.wikidata.org/wiki/Q1496376","display_name":"Gaussian process","level":3,"score":0.3319999873638153},{"id":"https://openalex.org/C186429297","wikidata":"https://www.wikidata.org/wiki/Q44451","display_name":"Hermite interpolation","level":3,"score":0.33160001039505005},{"id":"https://openalex.org/C2780551164","wikidata":"https://www.wikidata.org/wiki/Q2306599","display_name":"Column (typography)","level":3,"score":0.33149999380111694},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.3138999938964844},{"id":"https://openalex.org/C150899416","wikidata":"https://www.wikidata.org/wiki/Q1820378","display_name":"Transfer of learning","level":2,"score":0.31299999356269836},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.3043999969959259},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.290800005197525},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.2890999913215637},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2851000130176544},{"id":"https://openalex.org/C137800194","wikidata":"https://www.wikidata.org/wiki/Q11713455","display_name":"Interpolation (computer graphics)","level":3,"score":0.28459998965263367},{"id":"https://openalex.org/C118615104","wikidata":"https://www.wikidata.org/wiki/Q121416","display_name":"Discrete mathematics","level":1,"score":0.28360000252723694},{"id":"https://openalex.org/C61224824","wikidata":"https://www.wikidata.org/wiki/Q2260434","display_name":"Mixture model","level":2,"score":0.2696000039577484},{"id":"https://openalex.org/C201362023","wikidata":"https://www.wikidata.org/wiki/Q658574","display_name":"Hermite polynomials","level":2,"score":0.26420000195503235},{"id":"https://openalex.org/C166437778","wikidata":"https://www.wikidata.org/wiki/Q50695","display_name":"Quadratic function","level":3,"score":0.2614000141620636},{"id":"https://openalex.org/C129848803","wikidata":"https://www.wikidata.org/wiki/Q2564360","display_name":"Sample size determination","level":2,"score":0.2596000134944916},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2565999925136566}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2610.08420","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.08420","pdf_url":"https://arxiv.org/pdf/2610.08420","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.2610.08420","doi":"https://doi.org/10.48550/arxiv.2610.08420","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2610.08420","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:2610.08420","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.08420","pdf_url":"https://arxiv.org/pdf/2610.08420","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/G874360974","display_name":"DSGIANGO: The DynamicS of GradIent Descent AlGOrithm in Neural Networks","funder_award_id":"225837","funder_id":"https://openalex.org/F4320320924","funder_display_name":"Schweizerischer Nationalfonds zur F\u00f6rderung der Wissenschaftlichen Forschung"}],"funders":[{"id":"https://openalex.org/F4320320915","display_name":"\u00c9cole Polytechnique F\u00e9d\u00e9rale de Lausanne","ror":"https://ror.org/02s376052"},{"id":"https://openalex.org/F4320320924","display_name":"Schweizerischer Nationalfonds zur F\u00f6rderung der Wissenschaftlichen Forschung","ror":"https://ror.org/00yjd3n13"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7220908583.pdf","grobid_xml":"https://content.openalex.org/works/W7220908583.grobid-xml"},"referenced_works_count":9,"referenced_works":["https://openalex.org/W2083459869","https://openalex.org/W2121873815","https://openalex.org/W3020859587","https://openalex.org/W3157286395","https://openalex.org/W4250954493","https://openalex.org/W4403570080","https://openalex.org/W4404306248","https://openalex.org/W4412420048","https://openalex.org/W4417041451"],"related_works":["https://openalex.org/W7148405216","https://openalex.org/W4387356288","https://openalex.org/W4417308918","https://openalex.org/W7212065957","https://openalex.org/W7196951031","https://openalex.org/W4415671710","https://openalex.org/W7165784067","https://openalex.org/W7127606804","https://openalex.org/W4417095868","https://openalex.org/W3174629550"],"abstract_inverted_index":{"We":[0,13,35],"establish":[1],"a":[2,29,64,89,165,177],"polynomial":[3,92,166],"sample":[4,101,144,156],"complexity":[5,102],"separation":[6,134],"between":[7],"symmetry-aware":[8],"and":[9,24,53,71,110,147,162],"symmetry-agnostic":[10,122],"feature":[11],"learning.":[12],"study":[14],"growing-rank":[15],"multi-index":[16],"models":[17],"with":[18,77,85,94,143],"high-dimensional":[19],"Gaussian":[20],"covariates":[21],"in":[22,117,186],"$\\mathbb{R}^d$":[23],"$r=\u0398(d^\u03b4)$":[25],"teacher":[26,141,188],"directions":[27],"forming":[28],"cyclic":[30],"symmetry":[31,51],"orbit,":[32,189],"where":[33],"$0<\u03b4<1/2$.":[34],"compare":[36],"three":[37],"ways":[38],"of":[39,91,139,158],"exploiting":[40],"this":[41],"structure:":[42],"architectural":[43,159],"weight":[44,160],"sharing,":[45,161],"data":[46,79,152],"augmentation":[47,153],"over":[48,168],"the":[49,58,72,108,121,127,132,140,155,174,187,198],"full":[50],"group,":[52],"learning":[54],"without":[55,170],"access":[56],"to":[57,105,197],"symmetry.":[59,171],"In":[60],"particular,":[61],"we":[62,98],"analyze":[63],"symmetry-tied":[65],"convolutional":[66],"network,":[67,70],"an":[68],"untied":[69,74],"same":[73,133],"network":[75],"trained":[76],"full-group":[78,151],"augmentation,":[80],"using":[81],"spherical":[82],"online":[83],"SGD":[84],"correlation":[86],"loss.":[87],"For":[88,126,172],"class":[90],"links":[93],"information":[95],"exponent":[96],"$p\\ge3$,":[97,173],"prove":[99],"matching":[100],"bounds":[103],"up":[104],"logarithmic":[106],"factors:":[107],"tied":[109],"augmented":[111],"learners":[112],"achieve":[113],"weak":[114,137,199],"directional":[115],"recovery":[116,138,200],"$\\widetilde\u0398(d^{p-1})$":[118],"samples,":[119],"whereas":[120],"learner":[123],"requires":[124],"$\\widetilde\u0398(rd^{p-1})$.":[125],"pure":[128],"quadratic":[129],"Hermite":[130],"link,":[131],"holds":[135],"for":[136],"subspace,":[142],"complexities":[145],"$\\widetilde\u0398(d)$":[146],"$\\widetilde\u0398(rd)$,":[148],"respectively.":[149],"Thus,":[150],"matches":[154],"efficiency":[157],"both":[163],"provide":[164],"advantage":[167],"training":[169],"proof":[175],"reveals":[176],"two-stage":[178],"mechanism:":[179],"fluctuations":[180],"at":[181],"initialization":[182,208],"select":[183],"one":[184],"direction":[185],"after":[190],"which":[191],"localized":[192],"growth":[193],"amplifies":[194],"its":[195],"overlap":[196],"scale":[201],"while":[202],"competing":[203],"overlaps":[204],"remain":[205],"near":[206],"their":[207],"scale.":[209]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-10-09T00:00:00"}