{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:GOTL7QPMO5JD6BOEDZO364R7ZL","short_pith_number":"pith:GOTL7QPM","schema_version":"1.0","canonical_sha256":"33a6bfc1ec77523f05c41e5dbf723fcaedbd276f459b070020c693e4b5d8049f","source":{"kind":"arxiv","id":"2208.10967","version":5},"attestation_state":"computed","paper":{"title":"The Value of Out-of-Distribution Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Ashwin De Silva, Carey E. Priebe, Joshua T. Vogelstein, Pratik Chaudhari, Rahul Ramesh","submitted_at":"2022-08-23T13:41:01Z","abstract_excerpt":"We expect the generalization error to improve with more samples from a similar task, and to deteriorate with more samples from an out-of-distribution (OOD) task. In this work, we show a counter-intuitive phenomenon: the generalization error of a task can be a non-monotonic function of the number of OOD samples. As the number of OOD samples increases, the generalization error on the target task improves before deteriorating beyond a threshold. In other words, there is value in training on small amounts of OOD data. We use Fisher's Linear Discriminant on synthetic datasets and deep networks on c"},"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":"2208.10967","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-08-23T13:41:01Z","cross_cats_sorted":["cs.AI","cs.CV","stat.ML"],"title_canon_sha256":"8e45be14d62b913651a6cb3986f41d676d708338f1415a14c7873df33eeaa54e","abstract_canon_sha256":"98e749719c11330f1f9c37e4a2195b62b0552737afbe3dfb23961660cfac07c1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:32:34.180869Z","signature_b64":"2eaSejyQQWkMN7VGqcF+/suFrv2l9kAURyA0eBVDkOmdteTOrv1EkdR06vo5Xbwzyi1wcKTiSIT9dgtH/+qyCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"33a6bfc1ec77523f05c41e5dbf723fcaedbd276f459b070020c693e4b5d8049f","last_reissued_at":"2026-07-05T06:32:34.180341Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:32:34.180341Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Value of Out-of-Distribution Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Ashwin De Silva, Carey E. Priebe, Joshua T. Vogelstein, Pratik Chaudhari, Rahul Ramesh","submitted_at":"2022-08-23T13:41:01Z","abstract_excerpt":"We expect the generalization error to improve with more samples from a similar task, and to deteriorate with more samples from an out-of-distribution (OOD) task. In this work, we show a counter-intuitive phenomenon: the generalization error of a task can be a non-monotonic function of the number of OOD samples. As the number of OOD samples increases, the generalization error on the target task improves before deteriorating beyond a threshold. In other words, there is value in training on small amounts of OOD data. We use Fisher's Linear Discriminant on synthetic datasets and deep networks on c"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2208.10967","kind":"arxiv","version":5},"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/2208.10967/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":"2208.10967","created_at":"2026-07-05T06:32:34.180406+00:00"},{"alias_kind":"arxiv_version","alias_value":"2208.10967v5","created_at":"2026-07-05T06:32:34.180406+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2208.10967","created_at":"2026-07-05T06:32:34.180406+00:00"},{"alias_kind":"pith_short_12","alias_value":"GOTL7QPMO5JD","created_at":"2026-07-05T06:32:34.180406+00:00"},{"alias_kind":"pith_short_16","alias_value":"GOTL7QPMO5JD6BOE","created_at":"2026-07-05T06:32:34.180406+00:00"},{"alias_kind":"pith_short_8","alias_value":"GOTL7QPM","created_at":"2026-07-05T06:32:34.180406+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.12105","citing_title":"Out-of-distribution data supervision towards biomedical semantic segmentation","ref_index":6,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GOTL7QPMO5JD6BOEDZO364R7ZL","json":"https://pith.science/pith/GOTL7QPMO5JD6BOEDZO364R7ZL.json","graph_json":"https://pith.science/api/pith-number/GOTL7QPMO5JD6BOEDZO364R7ZL/graph.json","events_json":"https://pith.science/api/pith-number/GOTL7QPMO5JD6BOEDZO364R7ZL/events.json","paper":"https://pith.science/paper/GOTL7QPM"},"agent_actions":{"view_html":"https://pith.science/pith/GOTL7QPMO5JD6BOEDZO364R7ZL","download_json":"https://pith.science/pith/GOTL7QPMO5JD6BOEDZO364R7ZL.json","view_paper":"https://pith.science/paper/GOTL7QPM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2208.10967&json=true","fetch_graph":"https://pith.science/api/pith-number/GOTL7QPMO5JD6BOEDZO364R7ZL/graph.json","fetch_events":"https://pith.science/api/pith-number/GOTL7QPMO5JD6BOEDZO364R7ZL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GOTL7QPMO5JD6BOEDZO364R7ZL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GOTL7QPMO5JD6BOEDZO364R7ZL/action/storage_attestation","attest_author":"https://pith.science/pith/GOTL7QPMO5JD6BOEDZO364R7ZL/action/author_attestation","sign_citation":"https://pith.science/pith/GOTL7QPMO5JD6BOEDZO364R7ZL/action/citation_signature","submit_replication":"https://pith.science/pith/GOTL7QPMO5JD6BOEDZO364R7ZL/action/replication_record"}},"created_at":"2026-07-05T06:32:34.180406+00:00","updated_at":"2026-07-05T06:32:34.180406+00:00"}