{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:EKS6DQIMLDQFD6TMNR7C7LVAEM","short_pith_number":"pith:EKS6DQIM","schema_version":"1.0","canonical_sha256":"22a5e1c10c58e051fa6c6c7e2faea023245d5f91c3c6c44b536689a859a6d57d","source":{"kind":"arxiv","id":"2102.06571","version":3},"attestation_state":"computed","paper":{"title":"Bayesian Neural Network Priors Revisited","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Adri\\`a Garriga-Alonso, Florian Wenzel, Gunnar R\\\"atsch, Laurence Aitchison, Mark van der Wilk, Richard E. Turner, Sebastian W. Ober, Vincent Fortuin","submitted_at":"2021-02-12T15:18:06Z","abstract_excerpt":"Isotropic Gaussian priors are the de facto standard for modern Bayesian neural network inference. However, it is unclear whether these priors accurately reflect our true beliefs about the weight distributions or give optimal performance. To find better priors, we study summary statistics of neural network weights in networks trained using stochastic gradient descent (SGD). We find that convolutional neural network (CNN) and ResNet weights display strong spatial correlations, while fully connected networks (FCNNs) display heavy-tailed weight distributions. We show that building these observatio"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2102.06571","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2021-02-12T15:18:06Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"bbe43c070b97ef4fbe42ed2bfdc8717e83ec697d50ab77cfba0e30cc94119af8","abstract_canon_sha256":"1837be2d48c433629714423e15bb3c40db4e91578d36391971c66f6d415f5a4c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:05:30.350397Z","signature_b64":"k1/fw8SlcEuQCSOnx4UKsQ/GQdMeF06A3k+Z4icRS7S5tp9Msx2Pht/t65gCs1G9k5l9eyQ2WmtyIe7uZkyiCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"22a5e1c10c58e051fa6c6c7e2faea023245d5f91c3c6c44b536689a859a6d57d","last_reissued_at":"2026-07-05T04:05:30.349909Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:05:30.349909Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Bayesian Neural Network Priors Revisited","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Adri\\`a Garriga-Alonso, Florian Wenzel, Gunnar R\\\"atsch, Laurence Aitchison, Mark van der Wilk, Richard E. Turner, Sebastian W. Ober, Vincent Fortuin","submitted_at":"2021-02-12T15:18:06Z","abstract_excerpt":"Isotropic Gaussian priors are the de facto standard for modern Bayesian neural network inference. However, it is unclear whether these priors accurately reflect our true beliefs about the weight distributions or give optimal performance. To find better priors, we study summary statistics of neural network weights in networks trained using stochastic gradient descent (SGD). We find that convolutional neural network (CNN) and ResNet weights display strong spatial correlations, while fully connected networks (FCNNs) display heavy-tailed weight distributions. We show that building these observatio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2102.06571","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2102.06571/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2102.06571","created_at":"2026-07-05T04:05:30.349976+00:00"},{"alias_kind":"arxiv_version","alias_value":"2102.06571v3","created_at":"2026-07-05T04:05:30.349976+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2102.06571","created_at":"2026-07-05T04:05:30.349976+00:00"},{"alias_kind":"pith_short_12","alias_value":"EKS6DQIMLDQF","created_at":"2026-07-05T04:05:30.349976+00:00"},{"alias_kind":"pith_short_16","alias_value":"EKS6DQIMLDQFD6TM","created_at":"2026-07-05T04:05:30.349976+00:00"},{"alias_kind":"pith_short_8","alias_value":"EKS6DQIM","created_at":"2026-07-05T04:05:30.349976+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25745","citing_title":"Gaussian Mean Field Variational Inference can Overestimate Predictive Variance","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29448","citing_title":"Scalable Bayesian Spatial Mixture Modelling for Remote Sensing Image Segmentation","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13992","citing_title":"Physics-Informed Neural Networks for Methane Sorption: Cross-Gas Transfer Learning, Ensemble Collapse Under Physics Constraints, and Monte Carlo Dropout Uncertainty Quantification","ref_index":76,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EKS6DQIMLDQFD6TMNR7C7LVAEM","json":"https://pith.science/pith/EKS6DQIMLDQFD6TMNR7C7LVAEM.json","graph_json":"https://pith.science/api/pith-number/EKS6DQIMLDQFD6TMNR7C7LVAEM/graph.json","events_json":"https://pith.science/api/pith-number/EKS6DQIMLDQFD6TMNR7C7LVAEM/events.json","paper":"https://pith.science/paper/EKS6DQIM"},"agent_actions":{"view_html":"https://pith.science/pith/EKS6DQIMLDQFD6TMNR7C7LVAEM","download_json":"https://pith.science/pith/EKS6DQIMLDQFD6TMNR7C7LVAEM.json","view_paper":"https://pith.science/paper/EKS6DQIM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2102.06571&json=true","fetch_graph":"https://pith.science/api/pith-number/EKS6DQIMLDQFD6TMNR7C7LVAEM/graph.json","fetch_events":"https://pith.science/api/pith-number/EKS6DQIMLDQFD6TMNR7C7LVAEM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EKS6DQIMLDQFD6TMNR7C7LVAEM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EKS6DQIMLDQFD6TMNR7C7LVAEM/action/storage_attestation","attest_author":"https://pith.science/pith/EKS6DQIMLDQFD6TMNR7C7LVAEM/action/author_attestation","sign_citation":"https://pith.science/pith/EKS6DQIMLDQFD6TMNR7C7LVAEM/action/citation_signature","submit_replication":"https://pith.science/pith/EKS6DQIMLDQFD6TMNR7C7LVAEM/action/replication_record"}},"created_at":"2026-07-05T04:05:30.349976+00:00","updated_at":"2026-07-05T04:05:30.349976+00:00"}