{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:LWR33LOIGCYIBH3JHF2LZY5G35","short_pith_number":"pith:LWR33LOI","schema_version":"1.0","canonical_sha256":"5da3bdadc830b0809f693974bce3a6df7a2c44dc6bcfcd20a4f758c7e6885603","source":{"kind":"arxiv","id":"2412.17411","version":2},"attestation_state":"computed","paper":{"title":"Pretraining with random noise for uncertainty calibration","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.NE"],"primary_cat":"cs.LG","authors_text":"Jeonghwan Cheon, Se-Bum Paik","submitted_at":"2024-12-23T09:22:00Z","abstract_excerpt":"Uncertainty calibration is crucial for various machine learning applications, yet it remains challenging. Many models exhibit hallucinations - confident yet inaccurate responses - due to miscalibrated confidence. Here, we show that the common practice of random initialization in deep learning, often considered a standard technique, is an underlying cause of this miscalibration, leading to excessively high confidence in untrained networks. Our method, inspired by developmental neuroscience, addresses this issue by simply pretraining networks with random noise and labels, reducing overconfidence"},"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":"2412.17411","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2024-12-23T09:22:00Z","cross_cats_sorted":["cs.AI","cs.NE"],"title_canon_sha256":"34fba0c7c88ffddccefca78f6befc7d1afb1c609327043205b12b229fa40a6af","abstract_canon_sha256":"7b9f0df7469a8fcd5a2130d3c6547a3d44de74fe5126fbca640c5ce270103550"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:40:15.790260Z","signature_b64":"wA5qVzxDtRpCW8hJ4iklLw7YJqmN+pBduuaAHLKjJ7sdPwXALwHE14cc3w7zT4oAhazWOrgKdf8nGPunVF+rCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5da3bdadc830b0809f693974bce3a6df7a2c44dc6bcfcd20a4f758c7e6885603","last_reissued_at":"2026-07-05T10:40:15.789749Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:40:15.789749Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Pretraining with random noise for uncertainty calibration","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.NE"],"primary_cat":"cs.LG","authors_text":"Jeonghwan Cheon, Se-Bum Paik","submitted_at":"2024-12-23T09:22:00Z","abstract_excerpt":"Uncertainty calibration is crucial for various machine learning applications, yet it remains challenging. Many models exhibit hallucinations - confident yet inaccurate responses - due to miscalibrated confidence. Here, we show that the common practice of random initialization in deep learning, often considered a standard technique, is an underlying cause of this miscalibration, leading to excessively high confidence in untrained networks. Our method, inspired by developmental neuroscience, addresses this issue by simply pretraining networks with random noise and labels, reducing overconfidence"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.17411","kind":"arxiv","version":2},"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/2412.17411/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":"2412.17411","created_at":"2026-07-05T10:40:15.789810+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.17411v2","created_at":"2026-07-05T10:40:15.789810+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.17411","created_at":"2026-07-05T10:40:15.789810+00:00"},{"alias_kind":"pith_short_12","alias_value":"LWR33LOIGCYI","created_at":"2026-07-05T10:40:15.789810+00:00"},{"alias_kind":"pith_short_16","alias_value":"LWR33LOIGCYIBH3J","created_at":"2026-07-05T10:40:15.789810+00:00"},{"alias_kind":"pith_short_8","alias_value":"LWR33LOI","created_at":"2026-07-05T10:40:15.789810+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.20892","citing_title":"One-Time Soft Alignment Enables Resilient Learning without Weight Transport","ref_index":56,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LWR33LOIGCYIBH3JHF2LZY5G35","json":"https://pith.science/pith/LWR33LOIGCYIBH3JHF2LZY5G35.json","graph_json":"https://pith.science/api/pith-number/LWR33LOIGCYIBH3JHF2LZY5G35/graph.json","events_json":"https://pith.science/api/pith-number/LWR33LOIGCYIBH3JHF2LZY5G35/events.json","paper":"https://pith.science/paper/LWR33LOI"},"agent_actions":{"view_html":"https://pith.science/pith/LWR33LOIGCYIBH3JHF2LZY5G35","download_json":"https://pith.science/pith/LWR33LOIGCYIBH3JHF2LZY5G35.json","view_paper":"https://pith.science/paper/LWR33LOI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.17411&json=true","fetch_graph":"https://pith.science/api/pith-number/LWR33LOIGCYIBH3JHF2LZY5G35/graph.json","fetch_events":"https://pith.science/api/pith-number/LWR33LOIGCYIBH3JHF2LZY5G35/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LWR33LOIGCYIBH3JHF2LZY5G35/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LWR33LOIGCYIBH3JHF2LZY5G35/action/storage_attestation","attest_author":"https://pith.science/pith/LWR33LOIGCYIBH3JHF2LZY5G35/action/author_attestation","sign_citation":"https://pith.science/pith/LWR33LOIGCYIBH3JHF2LZY5G35/action/citation_signature","submit_replication":"https://pith.science/pith/LWR33LOIGCYIBH3JHF2LZY5G35/action/replication_record"}},"created_at":"2026-07-05T10:40:15.789810+00:00","updated_at":"2026-07-05T10:40:15.789810+00:00"}