{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:CE5TNMTPPD7STGNA5CVWFGRUTS","short_pith_number":"pith:CE5TNMTP","schema_version":"1.0","canonical_sha256":"113b36b26f78ff2999a0e8ab629a349c92287375a78b61dd4513fbe9c3fdab29","source":{"kind":"arxiv","id":"2006.11942","version":4},"attestation_state":"computed","paper":{"title":"Generalisation Guarantees for Continual Learning with Orthogonal Gradient Descent","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Masashi Sugiyama, Mehdi Abbana Bennani, Thang Doan","submitted_at":"2020-06-21T23:49:57Z","abstract_excerpt":"In Continual Learning settings, deep neural networks are prone to Catastrophic Forgetting. Orthogonal Gradient Descent was proposed to tackle the challenge. However, no theoretical guarantees have been proven yet. We present a theoretical framework to study Continual Learning algorithms in the Neural Tangent Kernel regime. This framework comprises closed form expression of the model through tasks and proxies for Transfer Learning, generalisation and tasks similarity. In this framework, we prove that OGD is robust to Catastrophic Forgetting then derive the first generalisation bound for SGD and"},"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":"2006.11942","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2020-06-21T23:49:57Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"b1f8566991167eeaafdf94526a11b304cf58e838e688aa0b8c4d7d09b9f967c5","abstract_canon_sha256":"de78f4d1b371a3abe03d6c16cd36b3a56f7ce4404f97e469d9e575f98f402732"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:56:55.434684Z","signature_b64":"Feko7KrYsN5rvmby54BcjH60saSDiXYGjKm0ZVsuv3fp/kNYdjHQ0wsd6w9bbZHi2QB6i4zOeYcycD4LWb4XBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"113b36b26f78ff2999a0e8ab629a349c92287375a78b61dd4513fbe9c3fdab29","last_reissued_at":"2026-07-05T01:56:55.434237Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:56:55.434237Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Generalisation Guarantees for Continual Learning with Orthogonal Gradient Descent","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Masashi Sugiyama, Mehdi Abbana Bennani, Thang Doan","submitted_at":"2020-06-21T23:49:57Z","abstract_excerpt":"In Continual Learning settings, deep neural networks are prone to Catastrophic Forgetting. Orthogonal Gradient Descent was proposed to tackle the challenge. However, no theoretical guarantees have been proven yet. We present a theoretical framework to study Continual Learning algorithms in the Neural Tangent Kernel regime. This framework comprises closed form expression of the model through tasks and proxies for Transfer Learning, generalisation and tasks similarity. In this framework, we prove that OGD is robust to Catastrophic Forgetting then derive the first generalisation bound for SGD and"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2006.11942","kind":"arxiv","version":4},"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/2006.11942/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":"2006.11942","created_at":"2026-07-05T01:56:55.434293+00:00"},{"alias_kind":"arxiv_version","alias_value":"2006.11942v4","created_at":"2026-07-05T01:56:55.434293+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2006.11942","created_at":"2026-07-05T01:56:55.434293+00:00"},{"alias_kind":"pith_short_12","alias_value":"CE5TNMTPPD7S","created_at":"2026-07-05T01:56:55.434293+00:00"},{"alias_kind":"pith_short_16","alias_value":"CE5TNMTPPD7STGNA","created_at":"2026-07-05T01:56:55.434293+00:00"},{"alias_kind":"pith_short_8","alias_value":"CE5TNMTP","created_at":"2026-07-05T01:56:55.434293+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18024","citing_title":"Catastrophic Forgetting is Low-Rank: A Function-Space Theory for Continual Adaptation","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27469","citing_title":"Architecture-driven Shift: towards a lightweight selector for capturing the trends of logit shift","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2603.11161","citing_title":"Algorithmic Task Capture, Computational Complexity, and Inductive Bias of Infinite Transformers","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12306","citing_title":"KAN-CL: Per-Knot Importance Regularization for Continual Learning with Kolmogorov-Arnold Networks","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07886","citing_title":"Characterizing and Correcting Effective Target Shift in Online Learning","ref_index":15,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CE5TNMTPPD7STGNA5CVWFGRUTS","json":"https://pith.science/pith/CE5TNMTPPD7STGNA5CVWFGRUTS.json","graph_json":"https://pith.science/api/pith-number/CE5TNMTPPD7STGNA5CVWFGRUTS/graph.json","events_json":"https://pith.science/api/pith-number/CE5TNMTPPD7STGNA5CVWFGRUTS/events.json","paper":"https://pith.science/paper/CE5TNMTP"},"agent_actions":{"view_html":"https://pith.science/pith/CE5TNMTPPD7STGNA5CVWFGRUTS","download_json":"https://pith.science/pith/CE5TNMTPPD7STGNA5CVWFGRUTS.json","view_paper":"https://pith.science/paper/CE5TNMTP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2006.11942&json=true","fetch_graph":"https://pith.science/api/pith-number/CE5TNMTPPD7STGNA5CVWFGRUTS/graph.json","fetch_events":"https://pith.science/api/pith-number/CE5TNMTPPD7STGNA5CVWFGRUTS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CE5TNMTPPD7STGNA5CVWFGRUTS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CE5TNMTPPD7STGNA5CVWFGRUTS/action/storage_attestation","attest_author":"https://pith.science/pith/CE5TNMTPPD7STGNA5CVWFGRUTS/action/author_attestation","sign_citation":"https://pith.science/pith/CE5TNMTPPD7STGNA5CVWFGRUTS/action/citation_signature","submit_replication":"https://pith.science/pith/CE5TNMTPPD7STGNA5CVWFGRUTS/action/replication_record"}},"created_at":"2026-07-05T01:56:55.434293+00:00","updated_at":"2026-07-05T01:56:55.434293+00:00"}