{"id":"https://openalex.org/W4415200437","doi":"https://doi.org/10.48550/arxiv.2505.10423","title":"The Power of Random Features and the Limits of Distribution-Free Gradient Descent","display_name":"The Power of Random Features and the Limits of Distribution-Free Gradient Descent","publication_year":2025,"publication_date":"2025-05-15","ids":{"openalex":"https://openalex.org/W4415200437","doi":"https://doi.org/10.48550/arxiv.2505.10423"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2505.10423","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.10423","pdf_url":"https://arxiv.org/pdf/2505.10423","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"parent_works":[],"lineage":["https://openalex.org/W4415200437"],"child_works_count":0,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2505.10423","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5020232766","display_name":"Ari Karchmer","orcid":"https://orcid.org/0000-0002-4692-4124","observed_orcids":["https://orcid.org/0000-0002-4692-4124"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Karchmer, Ari","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5069983576","display_name":"Eran Malach","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Malach, Eran","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":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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9932000041007996,"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":0.9932000041007996,"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":0.0027000000700354576,"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":0.0026000000070780516,"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/random-features","display_name":"random features","score":0.9570000171661377},{"id":"https://openalex.org/keywords/distribution-free-learning","display_name":"distribution-free learning","score":0.8560000061988831},{"id":"https://openalex.org/keywords/mini-batch-stochastic-gradient-descent","display_name":"mini-batch stochastic gradient descent","score":0.8930000066757202},{"id":"https://openalex.org/keywords/neural-network-training","display_name":"neural network training","score":0.6299999952316284}],"concepts":[{"id":"https://openalex.org/C33676613","wikidata":"https://www.wikidata.org/wiki/Q13415176","display_name":"Dimension (graph theory)","level":2,"score":0.7418000102043152},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.7020000219345093},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.6326000094413757},{"id":"https://openalex.org/C49937458","wikidata":"https://www.wikidata.org/wiki/Q2599292","display_name":"Probabilistic logic","level":2,"score":0.5867000222206116},{"id":"https://openalex.org/C117251300","wikidata":"https://www.wikidata.org/wiki/Q1849855","display_name":"Parametric statistics","level":2,"score":0.5443000197410583},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.5325999855995178},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.492900013923645},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.4860000014305115},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.4050000011920929},{"id":"https://openalex.org/C90119067","wikidata":"https://www.wikidata.org/wiki/Q43260","display_name":"Polynomial","level":2,"score":0.38530001044273376},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.3792000114917755},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.3781000077724457},{"id":"https://openalex.org/C163258240","wikidata":"https://www.wikidata.org/wiki/Q25342","display_name":"Power (physics)","level":2,"score":0.3662000000476837},{"id":"https://openalex.org/C115680565","wikidata":"https://www.wikidata.org/wiki/Q5977448","display_name":"Gradient method","level":2,"score":0.33329999446868896},{"id":"https://openalex.org/C122123141","wikidata":"https://www.wikidata.org/wiki/Q176623","display_name":"Random variable","level":2,"score":0.3280999958515167},{"id":"https://openalex.org/C21080849","wikidata":"https://www.wikidata.org/wiki/Q13611879","display_name":"Data point","level":2,"score":0.3093999922275543},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.30889999866485596},{"id":"https://openalex.org/C30732413","wikidata":"https://www.wikidata.org/wiki/Q17092636","display_name":"Intrinsic dimension","level":3,"score":0.3025999963283539},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.2992999851703644},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.26330000162124634},{"id":"https://openalex.org/C24574437","wikidata":"https://www.wikidata.org/wiki/Q7135228","display_name":"Parametric model","level":3,"score":0.262800008058548},{"id":"https://openalex.org/C8272713","wikidata":"https://www.wikidata.org/wiki/Q176737","display_name":"Stochastic process","level":2,"score":0.2624000012874603},{"id":"https://openalex.org/C55479107","wikidata":"https://www.wikidata.org/wiki/Q97663916","display_name":"Stochastic approximation","level":3,"score":0.2556999921798706}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2505.10423","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.10423","pdf_url":"https://arxiv.org/pdf/2505.10423","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2505.10423","doi":"https://doi.org/10.48550/arxiv.2505.10423","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.10423","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:2505.10423","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.10423","pdf_url":"https://arxiv.org/pdf/2505.10423","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4415200437.pdf","grobid_xml":"https://content.openalex.org/works/W4415200437.grobid-xml"},"referenced_works_count":4,"referenced_works":["https://openalex.org/W2894604724","https://openalex.org/W2982282910","https://openalex.org/W4386755420","https://openalex.org/W7133195721"],"related_works":["https://openalex.org/W7223017705","https://openalex.org/W2926417365","https://openalex.org/W2884243924","https://openalex.org/W4416246782","https://openalex.org/W4376603814","https://openalex.org/W7133224471","https://openalex.org/W4391158924","https://openalex.org/W4391673403","https://openalex.org/W3080149057","https://openalex.org/W2966363981"],"abstract_inverted_index":{"We":[0,139],"study":[1],"the":[2,43,50,71,81,112,129],"relationship":[3,146,154],"between":[4,160],"gradient-based":[5],"optimization":[6,14],"of":[7,15,18,61,66,73,89],"parametric":[8,28],"models":[9],"(e.g.,":[10],"neural":[11,93],"networks)":[12],"and":[13,76,151,162],"linear":[16],"combinations":[17],"random":[19,62],"features.":[20,63],"Our":[21],"main":[22],"result":[23],"shows":[24],"that":[25,141],"if":[26],"a":[27,58,117,144],"model":[29],"can":[30,53],"be":[31,55],"learned":[32],"using":[33,57],"mini-batch":[34],"stochastic":[35],"gradient":[36,74,97],"descent":[37],"(bSGD)":[38],"without":[39],"making":[40,101],"assumptions":[41,102],"about":[42,103],"data":[44,104],"distribution,":[45],"then":[46],"with":[47,147],"high":[48],"probability,":[49],"target":[51],"function":[52],"also":[54,115],"approximated":[56],"polynomial-sized":[59],"combination":[60,68],"The":[64],"size":[65],"this":[67,153],"depends":[69],"on":[70],"number":[72],"steps":[75],"numerical":[77],"precision":[78],"used":[79],"in":[80,92,109],"bSGD":[82],"process.":[83],"This":[84],"finding":[85],"reveals":[86],"fundamental":[87],"limitations":[88],"distribution-free":[90],"learning":[91],"networks":[94],"trained":[95],"by":[96,134],"descent,":[98],"highlighting":[99],"why":[100],"distributions":[105],"is":[106],"often":[107],"crucial":[108],"practice.":[110],"Along":[111],"way,":[113],"we":[114],"introduce":[116],"new":[118],"theoretical":[119],"framework":[120],"called":[121],"average":[122],"probabilistic":[123,130],"dimension":[124,131,164],"complexity":[125,132],"(adc),":[126],"which":[127],"extends":[128],"developed":[133],"Kamath":[135],"et":[136],"al.":[137],"(2020).":[138],"prove":[140],"adc":[142,161],"has":[143],"polynomial":[145],"statistical":[148],"query":[149],"dimension,":[150],"use":[152],"to":[155],"demonstrate":[156],"an":[157],"infinite":[158],"separation":[159],"standard":[163],"complexity.":[165]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2025-10-15T00:00:00"}