{"id":"https://openalex.org/W7160847294","doi":"https://doi.org/10.48550/arxiv.2605.07577","title":"Bilevel Graph Structure Learning, Revisited: Inner-Channel Origins of the Reported Gain","display_name":"Bilevel Graph Structure Learning, Revisited: Inner-Channel Origins of the Reported Gain","publication_year":2026,"publication_date":"2026-05-08","ids":{"openalex":"https://openalex.org/W7160847294","doi":"https://doi.org/10.48550/arxiv.2605.07577"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.07577","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07577","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,"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.07577","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5135849438","display_name":"Minkyoung Kim","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kim, Minkyoung","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5067151609","display_name":"Beakcheol Jang","orcid":"https://orcid.org/0000-0002-3911-5935"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jang, Beakcheol","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/T11273","display_name":"Advanced Graph Neural Networks","score":0.9865999817848206,"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.9865999817848206,"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":0.002400000113993883,"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/T10028","display_name":"Topic Modeling","score":0.0012000000569969416,"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/graph","display_name":"Graph","score":0.5730999708175659},{"id":"https://openalex.org/keywords/directed-graph","display_name":"Directed graph","score":0.3831000030040741},{"id":"https://openalex.org/keywords/bilevel-optimization","display_name":"Bilevel optimization","score":0.35370001196861267},{"id":"https://openalex.org/keywords/bernoullis-principle","display_name":"Bernoulli's principle","score":0.32670000195503235},{"id":"https://openalex.org/keywords/graph-theory","display_name":"Graph theory","score":0.3264000117778778},{"id":"https://openalex.org/keywords/graph-property","display_name":"Graph property","score":0.2957000136375427},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.2840999960899353}],"concepts":[{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.5730999708175659},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5561000108718872},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.43939998745918274},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4154999852180481},{"id":"https://openalex.org/C146380142","wikidata":"https://www.wikidata.org/wiki/Q1137726","display_name":"Directed graph","level":2,"score":0.3831000030040741},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.37689998745918274},{"id":"https://openalex.org/C3309286","wikidata":"https://www.wikidata.org/wiki/Q4907693","display_name":"Bilevel optimization","level":3,"score":0.35370001196861267},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.34540000557899475},{"id":"https://openalex.org/C152361515","wikidata":"https://www.wikidata.org/wiki/Q181328","display_name":"Bernoulli's principle","level":2,"score":0.32670000195503235},{"id":"https://openalex.org/C88230418","wikidata":"https://www.wikidata.org/wiki/Q131476","display_name":"Graph theory","level":2,"score":0.3264000117778778},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.3052000105381012},{"id":"https://openalex.org/C64339825","wikidata":"https://www.wikidata.org/wiki/Q722659","display_name":"Graph property","level":5,"score":0.2957000136375427},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.2840999960899353},{"id":"https://openalex.org/C148067565","wikidata":"https://www.wikidata.org/wiki/Q3323718","display_name":"Moral graph","level":5,"score":0.2754000127315521},{"id":"https://openalex.org/C27458966","wikidata":"https://www.wikidata.org/wiki/Q1187693","display_name":"Control flow graph","level":2,"score":0.2694999873638153},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.26249998807907104},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.2606000006198883},{"id":"https://openalex.org/C62611344","wikidata":"https://www.wikidata.org/wiki/Q1062658","display_name":"Node (physics)","level":2,"score":0.2590000033378601},{"id":"https://openalex.org/C48903430","wikidata":"https://www.wikidata.org/wiki/Q491370","display_name":"Graph partition","level":3,"score":0.25369998812675476},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.251800000667572}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.07577","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07577","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,"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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.07577","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07577","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,"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"},"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":{"Bilevel":[0],"graph":[1,9,20,69,95,99,155,159],"structure":[2,156],"learning":[3],"is":[4],"widely":[5],"understood":[6],"to":[7,28],"improve":[8],"neural":[10],"networks":[11],"by":[12],"jointly":[13],"optimizing":[14],"model":[15],"parameters":[16],"and":[17,93],"a":[18,52,64,94,130,150,162],"learned":[19],"structure,":[21],"with":[22,89,158],"the":[23,29,43,48,56,68,72,78,98,106,112,120,170,173],"resulting":[24],"performance":[25],"gain":[26,80,175],"attributed":[27],"rewired":[30],"adjacency.":[31],"We":[32,134,146],"find":[33],"that":[34,66,137],"this":[35],"attribution":[36],"may":[37],"be":[38],"overstated:":[39],"training-dynamics":[40],"effects":[41],"in":[42],"inner":[44,83,107],"loop,":[45],"rather":[46],"than":[47],"rewiring":[49,100],"itself,":[50],"capture":[51],"substantial":[53],"share":[54],"of":[55,85,97,119,172],"gain.":[57,145],"To":[58],"establish":[59],"this,":[60],"we":[61],"introduce":[62],"frozen-$\u03d5$,":[63],"control":[65],"freezes":[67],"while":[70],"retaining":[71],"inner-loop":[73],"training":[74,87],"schedule.":[75],"This":[76],"decomposes":[77],"bilevel":[79,114,154,174],"into":[81],"an":[82],"channel":[84,96,108],"$T$-step":[86],"dynamics":[88],"implicit":[90],"gradient":[91],"regularization":[92],"itself.":[101],"On":[102],"spatio-temporal":[103],"flow":[104],"forecasting":[105],"matches":[109],"or":[110],"exceeds":[111],"full":[113],"pipeline,":[115],"accounting":[116],"for":[117,127,153],"78-101%":[118],"gain;":[121],"on":[122,176],"node":[123],"classification":[124],"it":[125],"accounts":[126],"37-44%":[128],"under":[129],"Bernoulli":[131],"edge-level":[132],"parameterization.":[133],"also":[135],"verify":[136],"classical":[138],"spectral":[139],"diagnostics":[140],"can":[141],"dissociate":[142],"from":[143],"task":[144],"propose":[147],"frozen-$\u03d5$":[148],"as":[149,161],"standardized":[151],"diagnostic":[152],"learning,":[157],"distillation":[160],"method-agnostic":[163],"complement.":[164],"A":[165],"three-precondition":[166],"framework":[167],"further":[168],"predicts":[169],"sign":[171],"all":[177],"six":[178],"benchmarks.":[179]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-05-12T00:00:00"}