{"id":"https://openalex.org/W7157335535","doi":"https://doi.org/10.48550/arxiv.2604.23114","title":"A Tale of Two Variances: When Single-Seed Benchmarks Fail in Bayesian Deep Learning","display_name":"A Tale of Two Variances: When Single-Seed Benchmarks Fail in Bayesian Deep Learning","publication_year":2026,"publication_date":"2026-04-25","ids":{"openalex":"https://openalex.org/W7157335535","doi":"https://doi.org/10.48550/arxiv.2604.23114"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.23114","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.23114","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":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.23114","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5114176128","display_name":"Qishi Zhan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhan, Qishi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134775557","display_name":"Minxuan Hu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu, Minxuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134810655","display_name":"Liang He","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"He, Liang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134811164","display_name":"Guansu Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Guansu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5134784158","display_name":"Jiaxin Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Jiaxin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":5,"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/T12814","display_name":"Gaussian Processes and Bayesian Inference","score":0.426800012588501,"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/T12814","display_name":"Gaussian Processes and Bayesian Inference","score":0.426800012588501,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.20759999752044678,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.0551999993622303,"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/heteroscedasticity","display_name":"Heteroscedasticity","score":0.8306000232696533},{"id":"https://openalex.org/keywords/variance","display_name":"Variance (accounting)","score":0.6345999836921692},{"id":"https://openalex.org/keywords/metric","display_name":"Metric (unit)","score":0.5439000129699707},{"id":"https://openalex.org/keywords/dropout","display_name":"Dropout (neural networks)","score":0.47780001163482666},{"id":"https://openalex.org/keywords/bayes-theorem","display_name":"Bayes' theorem","score":0.4526999890804291},{"id":"https://openalex.org/keywords/regression","display_name":"Regression","score":0.4404999911785126},{"id":"https://openalex.org/keywords/trajectory","display_name":"Trajectory","score":0.4237000048160553},{"id":"https://openalex.org/keywords/point","display_name":"Point (geometry)","score":0.42329999804496765},{"id":"https://openalex.org/keywords/bayesian-probability","display_name":"Bayesian probability","score":0.4131999909877777}],"concepts":[{"id":"https://openalex.org/C101104100","wikidata":"https://www.wikidata.org/wiki/Q1063540","display_name":"Heteroscedasticity","level":2,"score":0.8306000232696533},{"id":"https://openalex.org/C196083921","wikidata":"https://www.wikidata.org/wiki/Q7915758","display_name":"Variance (accounting)","level":2,"score":0.6345999836921692},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.5877000093460083},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.5439000129699707},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.532800018787384},{"id":"https://openalex.org/C2776145597","wikidata":"https://www.wikidata.org/wiki/Q25339462","display_name":"Dropout (neural networks)","level":2,"score":0.47780001163482666},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4683000147342682},{"id":"https://openalex.org/C207201462","wikidata":"https://www.wikidata.org/wiki/Q182505","display_name":"Bayes' theorem","level":3,"score":0.4526999890804291},{"id":"https://openalex.org/C83546350","wikidata":"https://www.wikidata.org/wiki/Q1139051","display_name":"Regression","level":2,"score":0.4404999911785126},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4366999864578247},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.4237000048160553},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.42329999804496765},{"id":"https://openalex.org/C107673813","wikidata":"https://www.wikidata.org/wiki/Q812534","display_name":"Bayesian probability","level":2,"score":0.4131999909877777},{"id":"https://openalex.org/C41426520","wikidata":"https://www.wikidata.org/wiki/Q1192065","display_name":"Point estimation","level":2,"score":0.40139999985694885},{"id":"https://openalex.org/C139945424","wikidata":"https://www.wikidata.org/wiki/Q1940696","display_name":"Mean squared error","level":2,"score":0.396699994802475},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.3813000023365021},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.3765000104904175},{"id":"https://openalex.org/C189950617","wikidata":"https://www.wikidata.org/wiki/Q937228","display_name":"Property (philosophy)","level":2,"score":0.32089999318122864},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.31360000371932983},{"id":"https://openalex.org/C72169020","wikidata":"https://www.wikidata.org/wiki/Q194404","display_name":"Monotonic function","level":2,"score":0.3034999966621399},{"id":"https://openalex.org/C48921125","wikidata":"https://www.wikidata.org/wiki/Q10861030","display_name":"Linear regression","level":2,"score":0.2987000048160553},{"id":"https://openalex.org/C152877465","wikidata":"https://www.wikidata.org/wiki/Q208042","display_name":"Regression analysis","level":2,"score":0.29089999198913574},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.27410000562667847},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.27390000224113464},{"id":"https://openalex.org/C160234255","wikidata":"https://www.wikidata.org/wiki/Q812535","display_name":"Bayesian inference","level":3,"score":0.26579999923706055},{"id":"https://openalex.org/C5274069","wikidata":"https://www.wikidata.org/wiki/Q2285707","display_name":"Categorical variable","level":2,"score":0.263700008392334},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.26350000500679016},{"id":"https://openalex.org/C2779466056","wikidata":"https://www.wikidata.org/wiki/Q107630651","display_name":"Time point","level":2,"score":0.2630000114440918}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.23114","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.23114","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":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.23114","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.23114","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":"article"},"sustainable_development_goals":[{"display_name":"Peace, Justice and strong institutions","id":"https://metadata.un.org/sdg/16","score":0.6437786221504211}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"In":[0],"limited-data":[1],"settings,":[2],"a":[3,21,33,70,76,128,156],"single":[4],"endpoint":[5,217],"mean":[6,145],"of":[7,36,127,137,142,159,182],"an":[8],"evaluation":[9,222],"metric":[10],"such":[11],"as":[12,29],"the":[13,37,118,124,135,143,187,195,200,203,209],"Continuous":[14],"Ranked":[15],"Probability":[16],"Score":[17],"(CRPS)":[18],"is":[19,26],"itself":[20],"random":[22],"variable,":[23],"yet":[24],"it":[25,31],"routinely":[27],"reported":[28],"if":[30],"were":[32],"stable":[34],"property":[35],"method.":[38],"We":[39,149],"study":[40],"when":[41],"this":[42],"practice":[43],"fails.":[44],"Using":[45],"50":[46],"independent":[47],"repetitions":[48],"across":[49,61],"six":[50],"regression":[51],"datasets,":[52,98],"we":[53],"show":[54,107,150],"that":[55,151,202],"CRPS":[56,153],"variance":[57,79,90,109,119,154],"trajectories":[58],"differ":[59],"substantially":[60,193],"methods":[62],"and":[63,83,102,134,175,219],"are":[64],"not":[65],"always":[66],"well":[67],"described":[68],"by":[69,104],"smooth":[71,108],"power-law":[72],"decay.":[73],"Methods":[74],"with":[75,163,191,199],"learned":[77],"heteroscedastic":[78,189,204],"head,":[80],"namely":[81],"MAP":[82,130],"Deep":[84],"Ensembles,":[85],"can":[86],"develop":[87],"pronounced,":[88],"reproducible":[89],"peaks":[91,112],"at":[92,117],"intermediate":[93],"training":[94,205],"sizes":[95],"on":[96,121,168],"real":[97,170],"whereas":[99],"MC":[100],"Dropout":[101],"Bayes":[103],"Backprop":[105],"typically":[106],"contraction.":[110],"These":[111],"have":[113],"direct":[114,157],"practical":[115],"consequences:":[116],"peak":[120],"Seoul":[122],"Bike,":[123],"relative":[125],"RMSE":[126],"single-seed":[129,160],"estimate":[131],"reaches":[132],"93.6\\%,":[133],"probability":[136],"falling":[138],"within":[139],"\\(\\pm":[140],"10\\%\\)":[141],"repeated-run":[144],"drops":[146],"to":[147,208],"5.9\\%.":[148],"local":[152],"provides":[155],"signal":[158],"estimation":[161],"error,":[162],"Spearman":[164],"correlations":[165],"above":[166],"0.96":[167],"every":[169],"dataset.":[171],"Power-law":[172],"fit":[173],"quality":[174],"monotonicity":[176],"together":[177],"provide":[178],"compact":[179],"method-level":[180],"summaries":[181,215],"trajectory":[183,214],"regularity.":[184],"Finally,":[185],"replacing":[186],"standard":[188],"objective":[190,206],"\\(\u03b2\\)-NLL":[192],"reduces":[194],"irregular":[196],"behavior,":[197],"consistent":[198],"view":[201],"contributes":[207],"instability.":[210],"Practitioners":[211],"should":[212],"report":[213],"alongside":[216],"means":[218],"concentrate":[220],"repeated":[221],"in":[223],"high-variance":[224],"regions.":[225]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-04-29T00:00:00"}