{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:IPK5WID2E4CLBMD4QSSYVDJW7U","short_pith_number":"pith:IPK5WID2","schema_version":"1.0","canonical_sha256":"43d5db207a2704b0b07c84a58a8d36fd3f8bb5dc858ef4fa1ee92fa3d079fb63","source":{"kind":"arxiv","id":"2103.01133","version":3},"attestation_state":"computed","paper":{"title":"Posterior Meta-Replay for Continual Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Benjamin Ehret, Benjamin F. Grewe, Christian Henning, Francesco D'Angelo, Jo\\~ao Sacramento, Johannes von Oswald, Maria R. Cervera, Regina Traber, Seijin Kobayashi","submitted_at":"2021-03-01T17:08:35Z","abstract_excerpt":"Learning a sequence of tasks without access to i.i.d. observations is a widely studied form of continual learning (CL) that remains challenging. In principle, Bayesian learning directly applies to this setting, since recursive and one-off Bayesian updates yield the same result. In practice, however, recursive updating often leads to poor trade-off solutions across tasks because approximate inference is necessary for most models of interest. Here, we describe an alternative Bayesian approach where task-conditioned parameter distributions are continually inferred from data. We offer a practical "},"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":"2103.01133","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-03-01T17:08:35Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1a95340eb4a6a9bb788a1c7c3b7e22baa11c0ab4949c378eeb28bb2e683d57b6","abstract_canon_sha256":"006e9484c88533cf6c5041b734c0cfb590239a37214ad59fc2a912da53b3bbcb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:24:28.551776Z","signature_b64":"UVQE8GzmPBsTgSdtNBD9hynTS/eGPss3gBQfRvC3VAZXj9H1f4wcWuuBaVLN6PzH1wrsxjirnjAsCLgKjHZuBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"43d5db207a2704b0b07c84a58a8d36fd3f8bb5dc858ef4fa1ee92fa3d079fb63","last_reissued_at":"2026-07-05T03:24:28.551191Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:24:28.551191Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Posterior Meta-Replay for Continual Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Benjamin Ehret, Benjamin F. Grewe, Christian Henning, Francesco D'Angelo, Jo\\~ao Sacramento, Johannes von Oswald, Maria R. Cervera, Regina Traber, Seijin Kobayashi","submitted_at":"2021-03-01T17:08:35Z","abstract_excerpt":"Learning a sequence of tasks without access to i.i.d. observations is a widely studied form of continual learning (CL) that remains challenging. In principle, Bayesian learning directly applies to this setting, since recursive and one-off Bayesian updates yield the same result. In practice, however, recursive updating often leads to poor trade-off solutions across tasks because approximate inference is necessary for most models of interest. Here, we describe an alternative Bayesian approach where task-conditioned parameter distributions are continually inferred from data. We offer a practical "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2103.01133","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/2103.01133/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":"2103.01133","created_at":"2026-07-05T03:24:28.551264+00:00"},{"alias_kind":"arxiv_version","alias_value":"2103.01133v3","created_at":"2026-07-05T03:24:28.551264+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2103.01133","created_at":"2026-07-05T03:24:28.551264+00:00"},{"alias_kind":"pith_short_12","alias_value":"IPK5WID2E4CL","created_at":"2026-07-05T03:24:28.551264+00:00"},{"alias_kind":"pith_short_16","alias_value":"IPK5WID2E4CLBMD4","created_at":"2026-07-05T03:24:28.551264+00:00"},{"alias_kind":"pith_short_8","alias_value":"IPK5WID2","created_at":"2026-07-05T03:24:28.551264+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IPK5WID2E4CLBMD4QSSYVDJW7U","json":"https://pith.science/pith/IPK5WID2E4CLBMD4QSSYVDJW7U.json","graph_json":"https://pith.science/api/pith-number/IPK5WID2E4CLBMD4QSSYVDJW7U/graph.json","events_json":"https://pith.science/api/pith-number/IPK5WID2E4CLBMD4QSSYVDJW7U/events.json","paper":"https://pith.science/paper/IPK5WID2"},"agent_actions":{"view_html":"https://pith.science/pith/IPK5WID2E4CLBMD4QSSYVDJW7U","download_json":"https://pith.science/pith/IPK5WID2E4CLBMD4QSSYVDJW7U.json","view_paper":"https://pith.science/paper/IPK5WID2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2103.01133&json=true","fetch_graph":"https://pith.science/api/pith-number/IPK5WID2E4CLBMD4QSSYVDJW7U/graph.json","fetch_events":"https://pith.science/api/pith-number/IPK5WID2E4CLBMD4QSSYVDJW7U/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IPK5WID2E4CLBMD4QSSYVDJW7U/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IPK5WID2E4CLBMD4QSSYVDJW7U/action/storage_attestation","attest_author":"https://pith.science/pith/IPK5WID2E4CLBMD4QSSYVDJW7U/action/author_attestation","sign_citation":"https://pith.science/pith/IPK5WID2E4CLBMD4QSSYVDJW7U/action/citation_signature","submit_replication":"https://pith.science/pith/IPK5WID2E4CLBMD4QSSYVDJW7U/action/replication_record"}},"created_at":"2026-07-05T03:24:28.551264+00:00","updated_at":"2026-07-05T03:24:28.551264+00:00"}