{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2016:36TDEJBLTQDEC7J6XHUEDZM2L7","short_pith_number":"pith:36TDEJBL","schema_version":"1.0","canonical_sha256":"dfa632242b9c06417d3eb9e841e59a5feda7ec73d67e2ca31c4514df4bd60ce6","source":{"kind":"arxiv","id":"1612.00796","version":2},"attestation_state":"computed","paper":{"title":"Overcoming catastrophic forgetting in neural networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Agnieszka Grabska-Barwinska, Andrei A. Rusu, Claudia Clopath, Demis Hassabis, Dharshan Kumaran, Guillaume Desjardins, James Kirkpatrick, Joel Veness, John Quan, Kieran Milan, Neil Rabinowitz, Raia Hadsell, Razvan Pascanu, Tiago Ramalho","submitted_at":"2016-12-02T19:18:37Z","abstract_excerpt":"The ability to learn tasks in a sequential fashion is crucial to the development of artificial intelligence. Neural networks are not, in general, capable of this and it has been widely thought that catastrophic forgetting is an inevitable feature of connectionist models. We show that it is possible to overcome this limitation and train networks that can maintain expertise on tasks which they have not experienced for a long time. Our approach remembers old tasks by selectively slowing down learning on the weights important for those tasks. We demonstrate our approach is scalable and effective b"},"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":"1612.00796","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2016-12-02T19:18:37Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"ff6d5cf61325412bb64087b1cd26b4a5374d6413f5f2e48be4e9e66c54e8611c","abstract_canon_sha256":"9e4b49881e3ef6227b7ce1a2be6de5bef4739b9aadd2a79a133436fb6afe1c8a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:29:21.376638Z","signature_b64":"psNapiUoeA9BA3flKdsfOlUOHgKIIG+x6s7g5Zux/IOkEE2f01VsAYwXyqtYArrnvjf+pyrqg4xEjlIex26MBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"dfa632242b9c06417d3eb9e841e59a5feda7ec73d67e2ca31c4514df4bd60ce6","last_reissued_at":"2026-07-05T04:29:21.376161Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:29:21.376161Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Overcoming catastrophic forgetting in neural networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Agnieszka Grabska-Barwinska, Andrei A. Rusu, Claudia Clopath, Demis Hassabis, Dharshan Kumaran, Guillaume Desjardins, James Kirkpatrick, Joel Veness, John Quan, Kieran Milan, Neil Rabinowitz, Raia Hadsell, Razvan Pascanu, Tiago Ramalho","submitted_at":"2016-12-02T19:18:37Z","abstract_excerpt":"The ability to learn tasks in a sequential fashion is crucial to the development of artificial intelligence. Neural networks are not, in general, capable of this and it has been widely thought that catastrophic forgetting is an inevitable feature of connectionist models. We show that it is possible to overcome this limitation and train networks that can maintain expertise on tasks which they have not experienced for a long time. Our approach remembers old tasks by selectively slowing down learning on the weights important for those tasks. We demonstrate our approach is scalable and effective b"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1612.00796","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/1612.00796/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":"1612.00796","created_at":"2026-07-05T04:29:21.376217+00:00"},{"alias_kind":"arxiv_version","alias_value":"1612.00796v2","created_at":"2026-07-05T04:29:21.376217+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1612.00796","created_at":"2026-07-05T04:29:21.376217+00:00"},{"alias_kind":"pith_short_12","alias_value":"36TDEJBLTQDE","created_at":"2026-07-05T04:29:21.376217+00:00"},{"alias_kind":"pith_short_16","alias_value":"36TDEJBLTQDEC7J6","created_at":"2026-07-05T04:29:21.376217+00:00"},{"alias_kind":"pith_short_8","alias_value":"36TDEJBL","created_at":"2026-07-05T04:29:21.376217+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":13,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26050","citing_title":"Natural Ungrokking: Asymmetric Control of Which Rules Survive Pretraining","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.24747","citing_title":"Scaling Laws for Task-Specific LLM Distillation","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11712","citing_title":"Substrate Asymmetry in User-Side Memory: A Diagnostic Framework","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06494","citing_title":"TailLoR: Protecting Principal Components in Parameter-Efficient Continual Learning","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01967","citing_title":"Training Prompt Matters: State-Adaptive Optimization for Robust Fine-Tuning","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02398","citing_title":"A Local Perturbation Theory for Cross-Domain Interference and Recovery in Multi-Domain RL","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00132","citing_title":"Foundation-Preserving Adaptation via Generalized Rayleigh-Quotient Optimization","ref_index":59,"is_internal_anchor":false},{"citing_arxiv_id":"2505.16831","citing_title":"Unlearning Isn't Deletion: Investigating Reversibility of Machine Unlearning in LLMs","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20296","citing_title":"Spectral Unforgetting: Post-Hoc Recovery of Damaged Capabilities Without Retraining","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19080","citing_title":"MANGO: Meta-Adaptive Network Gradient Optimization for Online Continual Learning","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19890","citing_title":"GoTTA be Diverse: Rethinking Memory Policies for Test-Time Adaptation","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11042","citing_title":"Improving Layout Representation Learning Across Inconsistently Annotated Datasets via Agentic Harmonization","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06690","citing_title":"State Representation and Termination for Recursive Reasoning Systems","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/36TDEJBLTQDEC7J6XHUEDZM2L7","json":"https://pith.science/pith/36TDEJBLTQDEC7J6XHUEDZM2L7.json","graph_json":"https://pith.science/api/pith-number/36TDEJBLTQDEC7J6XHUEDZM2L7/graph.json","events_json":"https://pith.science/api/pith-number/36TDEJBLTQDEC7J6XHUEDZM2L7/events.json","paper":"https://pith.science/paper/36TDEJBL"},"agent_actions":{"view_html":"https://pith.science/pith/36TDEJBLTQDEC7J6XHUEDZM2L7","download_json":"https://pith.science/pith/36TDEJBLTQDEC7J6XHUEDZM2L7.json","view_paper":"https://pith.science/paper/36TDEJBL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1612.00796&json=true","fetch_graph":"https://pith.science/api/pith-number/36TDEJBLTQDEC7J6XHUEDZM2L7/graph.json","fetch_events":"https://pith.science/api/pith-number/36TDEJBLTQDEC7J6XHUEDZM2L7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/36TDEJBLTQDEC7J6XHUEDZM2L7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/36TDEJBLTQDEC7J6XHUEDZM2L7/action/storage_attestation","attest_author":"https://pith.science/pith/36TDEJBLTQDEC7J6XHUEDZM2L7/action/author_attestation","sign_citation":"https://pith.science/pith/36TDEJBLTQDEC7J6XHUEDZM2L7/action/citation_signature","submit_replication":"https://pith.science/pith/36TDEJBLTQDEC7J6XHUEDZM2L7/action/replication_record"}},"created_at":"2026-07-05T04:29:21.376217+00:00","updated_at":"2026-07-05T04:29:21.376217+00:00"}