{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:RGNXWHTE3BEAAUQZOJXUIPXQ57","short_pith_number":"pith:RGNXWHTE","schema_version":"1.0","canonical_sha256":"899b7b1e64d848005219726f443ef0efee3a88015c3c07a97fe8c85e576ff41e","source":{"kind":"arxiv","id":"2402.18762","version":1},"attestation_state":"computed","paper":{"title":"Disentangling the Causes of Plasticity Loss in Neural Networks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Clare Lyle, Hado van Hasselt, James Martens, Khimya Khetarpal, Razvan Pascanu, Will Dabney, Zeyu Zheng","submitted_at":"2024-02-29T00:02:33Z","abstract_excerpt":"Underpinning the past decades of work on the design, initialization, and optimization of neural networks is a seemingly innocuous assumption: that the network is trained on a \\textit{stationary} data distribution. In settings where this assumption is violated, e.g.\\ deep reinforcement learning, learning algorithms become unstable and brittle with respect to hyperparameters and even random seeds. One factor driving this instability is the loss of plasticity, meaning that updating the network's predictions in response to new information becomes more difficult as training progresses. While many r"},"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":"2402.18762","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-29T00:02:33Z","cross_cats_sorted":[],"title_canon_sha256":"fc20fcae42328a671966f9c5fca1934e53319574272c3a0ef9be96b1bfbba89e","abstract_canon_sha256":"9b099b952583d89629b44cd922f3fc7f6fd0b2b30000a3b3d1946a8b3ab1bd80"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:50:27.607356Z","signature_b64":"qmnIPfTdrB5o7EXSXH1jZRZ58lbOiThlIuamJSFyq7D+xmsobc6GHysWQzs+aFrk0CuI1Er4cm+GeJGjGUeaBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"899b7b1e64d848005219726f443ef0efee3a88015c3c07a97fe8c85e576ff41e","last_reissued_at":"2026-07-05T07:50:27.606845Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:50:27.606845Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Disentangling the Causes of Plasticity Loss in Neural Networks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Clare Lyle, Hado van Hasselt, James Martens, Khimya Khetarpal, Razvan Pascanu, Will Dabney, Zeyu Zheng","submitted_at":"2024-02-29T00:02:33Z","abstract_excerpt":"Underpinning the past decades of work on the design, initialization, and optimization of neural networks is a seemingly innocuous assumption: that the network is trained on a \\textit{stationary} data distribution. In settings where this assumption is violated, e.g.\\ deep reinforcement learning, learning algorithms become unstable and brittle with respect to hyperparameters and even random seeds. One factor driving this instability is the loss of plasticity, meaning that updating the network's predictions in response to new information becomes more difficult as training progresses. While many r"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.18762","kind":"arxiv","version":1},"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/2402.18762/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":"2402.18762","created_at":"2026-07-05T07:50:27.606909+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.18762v1","created_at":"2026-07-05T07:50:27.606909+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.18762","created_at":"2026-07-05T07:50:27.606909+00:00"},{"alias_kind":"pith_short_12","alias_value":"RGNXWHTE3BEA","created_at":"2026-07-05T07:50:27.606909+00:00"},{"alias_kind":"pith_short_16","alias_value":"RGNXWHTE3BEAAUQZ","created_at":"2026-07-05T07:50:27.606909+00:00"},{"alias_kind":"pith_short_8","alias_value":"RGNXWHTE","created_at":"2026-07-05T07:50:27.606909+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":11,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25527","citing_title":"Beyond One-Size-Fits-All: Diagnosis-Driven Online Reinforcement Learning with Offline Priors","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09762","citing_title":"Preserving Plasticity in Continual Learning via Dynamical Isometry","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06418","citing_title":"Double Preconditioning (DoPr): Optimization for Test-Time Performance, not Validation Loss","ref_index":78,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03382","citing_title":"Local Guidance, Global Impact: Gaussian-Reshaped Trust Region Unlocks Behavior Transitions","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26357","citing_title":"Balancing Plasticity and Stability with Fast and Slow Successor Features","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23565","citing_title":"Understanding Goal Generalisation in Sequential Reinforcement Learning","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2411.04832","citing_title":"Plasticity Loss in Deep Reinforcement Learning: A Survey","ref_index":68,"is_internal_anchor":false},{"citing_arxiv_id":"2509.22562","citing_title":"Activation Function Design Sustains Plasticity in Continual Learning","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09028","citing_title":"Plasticity-Enhanced Multi-Agent Mixture of Experts for Dynamic Objective Adaptation in UAVs-Assisted Emergency Communication Networks","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18857","citing_title":"Task Switching Without Forgetting via Proximal Decoupling","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15414","citing_title":"Beyond Single-Model Optimization: Preserving Plasticity in Continual Reinforcement Learning","ref_index":20,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RGNXWHTE3BEAAUQZOJXUIPXQ57","json":"https://pith.science/pith/RGNXWHTE3BEAAUQZOJXUIPXQ57.json","graph_json":"https://pith.science/api/pith-number/RGNXWHTE3BEAAUQZOJXUIPXQ57/graph.json","events_json":"https://pith.science/api/pith-number/RGNXWHTE3BEAAUQZOJXUIPXQ57/events.json","paper":"https://pith.science/paper/RGNXWHTE"},"agent_actions":{"view_html":"https://pith.science/pith/RGNXWHTE3BEAAUQZOJXUIPXQ57","download_json":"https://pith.science/pith/RGNXWHTE3BEAAUQZOJXUIPXQ57.json","view_paper":"https://pith.science/paper/RGNXWHTE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.18762&json=true","fetch_graph":"https://pith.science/api/pith-number/RGNXWHTE3BEAAUQZOJXUIPXQ57/graph.json","fetch_events":"https://pith.science/api/pith-number/RGNXWHTE3BEAAUQZOJXUIPXQ57/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RGNXWHTE3BEAAUQZOJXUIPXQ57/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RGNXWHTE3BEAAUQZOJXUIPXQ57/action/storage_attestation","attest_author":"https://pith.science/pith/RGNXWHTE3BEAAUQZOJXUIPXQ57/action/author_attestation","sign_citation":"https://pith.science/pith/RGNXWHTE3BEAAUQZOJXUIPXQ57/action/citation_signature","submit_replication":"https://pith.science/pith/RGNXWHTE3BEAAUQZOJXUIPXQ57/action/replication_record"}},"created_at":"2026-07-05T07:50:27.606909+00:00","updated_at":"2026-07-05T07:50:27.606909+00:00"}